# Cloud Security Risk Assessment: every page as markdown A 40-question risk assessment of AWS, Azure and Google Cloud access: a score per domain, five fixes, and a PDF and XLSX export made in your browser. --- # Cloud security risk assessment: score your cloud access layer in 20 minutes Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The cloud security risk assessment asks 40 questions about the cloud access layer of AWS, Azure and Google Cloud, in six domains, and takes about 20 minutes. It returns a score per domain, the five fixes that matter most, and a PDF and XLSX export made in your browser. Answers never leave the page. 1. Does the root user (or Global Administrator, or Organization Admin) of every account have phishing-resistant MFA and no access keys? (CIS AWS Foundations Benchmark v5.0 1.4 and 1.5) 2. Do all people reach the cloud through central single sign-on with MFA, with no local IAM users or standalone cloud passwords? (Microsoft cloud security benchmark IM-6) 3. Are long-lived access keys and service account keys absent, or rotated at least every 90 days and unused ones deleted? (Google Cloud security best practices Service account keys) 4. Do workloads get cloud access through attached roles, managed identities or workload identity federation instead of stored secrets? (Microsoft cloud security benchmark IM-3) 5. Is the number of people with administrator-level roles small, named and separate from their daily accounts? (Microsoft cloud security benchmark PA-1) 6. Are privileged roles granted only for a limited time, on request, instead of held permanently? (Microsoft cloud security benchmark PA-2) 7. Are permissions reviewed at least every 90 days and unused ones removed? (Microsoft cloud security benchmark PA-4) 8. Are emergency (break-glass) accounts in place, protected, monitored and tested at least twice a year? (Microsoft cloud security benchmark PA-5) 9. Is every SSH, RDP and database port closed to the whole internet (0.0.0.0/0 and ::/0)? (Google Cloud security best practices VPC firewall rules) 10. Do your data stores and platform services (storage, databases, secrets, AI) accept traffic only through private endpoints? (Microsoft cloud security benchmark NS-2) 11. Do new compute instances and databases get no public IP address by default? (Microsoft cloud security benchmark NS-1) 12. Do administrators reach servers through a brokered, logged service (Session Manager, Bastion, Identity-Aware Proxy) with no inbound admin ports? (Microsoft cloud security benchmark PA-6) 13. Is the instance metadata service hardened (IMDSv2 required on AWS, managed identity endpoint protected on Azure, metadata concealment on GKE)? (AWS documentation Instance metadata) 14. Is outbound traffic from production workloads filtered to known destinations? (Microsoft cloud security benchmark NS-3) 15. Do you analyze which resources are reachable from the internet and from other networks (peering, transit, cross-account) at least monthly? (Microsoft cloud security benchmark NS-7) 16. Is public access to object storage blocked at the organization level, not just per bucket? (AWS documentation S3 Block Public Access) 17. Do your resource policies and sharing rules limit access to identities in your own organization? (AWS documentation Data perimeter) 18. Can your most sensitive data be read only from trusted networks or endpoints (VPC endpoints, service perimeters, private links)? (Google Cloud security best practices VPC Service Controls) 19. Do you review findings about resources shared outside your organization at least weekly? (AWS documentation IAM Access Analyzer) 20. Are sensitive data encrypted with customer-managed keys whose use is limited to named identities and services? (Microsoft cloud security benchmark DP-5) 21. Do you know which storage locations hold regulated or sensitive data, and is that list up to date? (Microsoft cloud security benchmark DP-1) 22. Is every third-party and cross-account role protected by an external ID or condition, with an owner and an end date? (AWS documentation IAM best practices) 23. Does each AI agent, pipeline and application have its own cloud identity, not a shared one or a person's credentials? (Microsoft cloud security benchmark AI-4) 24. Is the list of actions each AI agent can perform limited to what its task needs, and written down? (Microsoft cloud security benchmark AI-4) 25. Can only approved models and AI services be used in your accounts, enforced by policy? (Microsoft cloud security benchmark AI-1) 26. Is the data an AI system can read (retrieval stores, training data, tool outputs) limited to what its users may see? (Google Cloud security best practices AI and ML security) 27. Do destructive or high-impact agent actions (delete, grant access, pay, send externally) need a human approval? (Microsoft cloud security benchmark AI-5) 28. Are model calls, tool calls and the identity behind each logged and kept? (Microsoft cloud security benchmark AI-6) 29. Do organization-level policies block the riskiest actions before they happen (SCPs, Azure Policy deny effects, organization policy constraints)? (AWS documentation Service control policies) 30. Do infrastructure changes go through code review and pipelines, with console changes in production flagged? (Microsoft cloud security benchmark PV-1) 31. Does a continuous check compare live configuration against your rules and report differences within a day? (AWS documentation AWS Config) 32. Do you test a new guardrail in audit or report-only mode and see what it would block before you enforce it? (Microsoft documentation Azure Policy) 33. Does every guardrail exception have an owner, a reason and an expiry date? (Microsoft cloud security benchmark PV-2) 34. Can administrators inside a workload account not turn off logging, guardrails or security services? (Microsoft cloud security benchmark PV-2) 35. Are control-plane audit logs on in every account, subscription and project, in every region, including data access logs for sensitive stores? (Microsoft cloud security benchmark LT-3) 36. Are audit logs stored in a separate, restricted account where workload administrators cannot change or delete them? (Microsoft cloud security benchmark LT-5) 37. Do you keep audit logs for at least 365 days, with the most recent 90 days searchable? (Microsoft cloud security benchmark LT-6) 38. Do alerts fire, and reach a person, for root or global administrator sign-ins, policy changes, new keys and disabled logging? (Microsoft cloud security benchmark LT-2) 39. Within one working day, can you produce a list of who can access a given sensitive data store and how they got that access? (Microsoft cloud security benchmark PA-4) 40. Have you practiced cutting off all access of one compromised identity within one hour, and written down the steps? (NIST SP 800-53 Rev. 5 IR-4) ## The six domains | Domain | Weight | Questions | Why it is weighted | | --- | --- | --- | --- | | Identities and entitlements | 25 | 8 | People and workloads that hold standing power over the cloud account are the most common route from one stolen credential to a full breach. | | Network reachability | 20 | 7 | Reachability decides which of your services an attacker can touch before any credential is needed. | | Data perimeter | 20 | 7 | A data perimeter limits who can reach your data and from where, even when an identity or a policy is wrong. | | AI workloads | 15 | 6 | AI services and agents act with cloud credentials, so what they can call and read is an access question. | | Guardrails and drift | 10 | 6 | Preventive controls stop a bad change before it lands, and drift checks catch the ones that slip past. | | Evidence | 10 | 6 | Logs and records are what let you prove who had access, detect misuse and respond. | Workload vulnerabilities and posture scanning are outside this assessment: it scores who and what can reach your cloud resources, not the software running on them. ## Other ways to assess The [directory of assessment options](/options/) lists 15 provider tools, open source scanners and vendor-run assessments by the same fields, in alphabetical order. ## Sources - CIS AWS Foundations Benchmark v5.0 1.4 and 1.5: https://www.cisecurity.org/benchmark/amazon_web_services (read 16 Nov 2026) - Microsoft cloud security benchmark IM-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - Google Cloud security best practices Service account keys: https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) - Microsoft cloud security benchmark IM-3: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - Microsoft cloud security benchmark PA-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - Microsoft cloud security benchmark PA-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/ --- # The 40-question assessment: how the cloud access score is calculated Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The method: 40 questions in six domains, answers worth 1, 0.5 or 0, weights from 1 to 3, a domain score from 0 to 100 and five fixes chosen by the largest weighted gap, with ties broken by lower effort. This page shows the arithmetic, the domain weights and the sources for the controls. Each of the 40 questions has three answers: yes in every account, in some accounts, and no or not known. They are worth 1, 0.5 and 0. A question carries a weight of 1, 2 or 3 that records how much a wrong answer moves risk. The 40 weights add up to 84 and 12 questions weigh 3. A domain score is the sum of value times weight over the domain's questions, divided by the sum of weights, times 100. The six domains hold 8, 7, 7, 6, 6 and 6 questions. Domain weights of 25, 20, 20, 15, 10 and 10, which add up to 100, apply only to an optional overall figure. Unknown counts as no, so a missing control and an unprovable one score the same. The five fixes are the questions with the largest weight times one minus the answer value. Ties go to the lower fix effort, rated low, medium or high, and then to question order. Every question names the control it comes from, and every question page gives a test for AWS, Azure and Google Cloud, so a score can be checked rather than trusted. ## What each answer is worth | Question | Domain | Weight | Answers and values | Fix effort | | --- | --- | --- | --- | --- | | [1. Does the root user (or Global Administrator, or Organization Admin) of every account have phishing-resistant MFA and no access keys?](/questions/root-account-mfa/) | Identities and entitlements | 3 | Yes, in every account = 1; In some accounts = 0.5; No or not known = 0 | low | | [2. Do all people reach the cloud through central single sign-on with MFA, with no local IAM users or standalone cloud passwords?](/questions/workforce-sso-mfa/) | Identities and entitlements | 3 | Yes, all people use SSO with MFA = 1; Mostly, with some local users = 0.5; No or not known = 0 | medium | | [3. Are long-lived access keys and service account keys absent, or rotated at least every 90 days and unused ones deleted?](/questions/long-lived-keys/) | Identities and entitlements | 3 | Yes, none or all under 90 days = 1; Some are older or unused = 0.5; No or not known = 0 | medium | | [4. Do workloads get cloud access through attached roles, managed identities or workload identity federation instead of stored secrets?](/questions/workload-identity/) | Identities and entitlements | 2 | Yes, for all workloads = 1; For some workloads = 0.5; No or not known = 0 | high | | [5. Is the number of people with administrator-level roles small, named and separate from their daily accounts?](/questions/admin-roles-limited/) | Identities and entitlements | 2 | Yes, small, named and separate = 1; Small but not separate = 0.5; No or not known = 0 | low | | [6. Are privileged roles granted only for a limited time, on request, instead of held permanently?](/questions/no-standing-privilege/) | Identities and entitlements | 3 | Yes, for all privileged roles = 1; For some privileged roles = 0.5; No or not known = 0 | high | | [7. Are permissions reviewed at least every 90 days and unused ones removed?](/questions/access-reviews/) | Identities and entitlements | 2 | Yes, reviewed and removed = 1; Reviewed, but removal is not enforced = 0.5; No or not known = 0 | medium | | [8. Are emergency (break-glass) accounts in place, protected, monitored and tested at least twice a year?](/questions/emergency-access/) | Identities and entitlements | 1 | Yes, protected, alerted and tested = 1; Exist but untested or unmonitored = 0.5; No or not known = 0 | low | | [9. Is every SSH, RDP and database port closed to the whole internet (0.0.0.0/0 and ::/0)?](/questions/no-open-admin-ports/) | Network reachability | 3 | Yes, none are open = 1; Open on a few exceptions = 0.5; No or not known = 0 | low | | [10. Do your data stores and platform services (storage, databases, secrets, AI) accept traffic only through private endpoints?](/questions/private-service-endpoints/) | Network reachability | 2 | Yes, for all production services = 1; For some services = 0.5; No or not known = 0 | high | | [11. Do new compute instances and databases get no public IP address by default?](/questions/no-default-public-ip/) | Network reachability | 2 | Yes, enforced by policy = 1; By convention, not enforced = 0.5; No or not known = 0 | medium | | [12. Do administrators reach servers through a brokered, logged service (Session Manager, Bastion, Identity-Aware Proxy) with no inbound admin ports?](/questions/brokered-admin-access/) | Network reachability | 2 | Yes, for all servers = 1; For some servers = 0.5; No or not known = 0 | medium | | [13. Is the instance metadata service hardened (IMDSv2 required on AWS, managed identity endpoint protected on Azure, metadata concealment on GKE)?](/questions/metadata-service-hardened/) | Network reachability | 3 | Yes, enforced everywhere = 1; On new resources only = 0.5; No or not known = 0 | low | | [14. Is outbound traffic from production workloads filtered to known destinations?](/questions/egress-controlled/) | Network reachability | 1 | Yes, deny by default with an allow list = 1; Partly, for some networks = 0.5; No or not known = 0 | high | | [15. Do you analyze which resources are reachable from the internet and from other networks (peering, transit, cross-account) at least monthly?](/questions/reachability-analysis/) | Network reachability | 1 | Yes, on a schedule with owners = 1; Ad hoc = 0.5; No or not known = 0 | medium | | [16. Is public access to object storage blocked at the organization level, not just per bucket?](/questions/public-storage-blocked/) | Data perimeter | 3 | Yes, organization-wide = 1; On most accounts or buckets = 0.5; No or not known = 0 | low | | [17. Do your resource policies and sharing rules limit access to identities in your own organization?](/questions/organization-boundary/) | Data perimeter | 3 | Yes, enforced by policy = 1; In some policies only = 0.5; No or not known = 0 | medium | | [18. Can your most sensitive data be read only from trusted networks or endpoints (VPC endpoints, service perimeters, private links)?](/questions/trusted-network-condition/) | Data perimeter | 2 | Yes, for all sensitive data = 1; For some data stores = 0.5; No or not known = 0 | high | | [19. Do you review findings about resources shared outside your organization at least weekly?](/questions/external-access-findings/) | Data perimeter | 2 | Yes, weekly or continuous with owners = 1; Occasionally = 0.5; No or not known = 0 | low | | [20. Are sensitive data encrypted with customer-managed keys whose use is limited to named identities and services?](/questions/key-control/) | Data perimeter | 2 | Yes, for sensitive data = 1; For some data = 0.5; No or not known = 0 | medium | | [21. Do you know which storage locations hold regulated or sensitive data, and is that list up to date?](/questions/sensitive-data-located/) | Data perimeter | 1 | Yes, scanned and tagged = 1; From memory or partly = 0.5; No or not known = 0 | medium | | [22. Is every third-party and cross-account role protected by an external ID or condition, with an owner and an end date?](/questions/third-party-access/) | Data perimeter | 2 | Yes, all have conditions, owners and dates = 1; Some do = 0.5; No or not known = 0 | medium | | [23. Does each AI agent, pipeline and application have its own cloud identity, not a shared one or a person's credentials?](/questions/ai-own-identity/) | AI workloads | 3 | Yes, one identity per agent or application = 1; Some share an identity = 0.5; No or not known = 0 | medium | | [24. Is the list of actions each AI agent can perform limited to what its task needs, and written down?](/questions/ai-tool-permissions/) | AI workloads | 3 | Yes, scoped and reviewed = 1; Partly scoped = 0.5; No or not known = 0 | high | | [25. Can only approved models and AI services be used in your accounts, enforced by policy?](/questions/approved-models-only/) | AI workloads | 2 | Yes, enforced by policy = 1; By guidance only = 0.5; No or not known = 0 | low | | [26. Is the data an AI system can read (retrieval stores, training data, tool outputs) limited to what its users may see?](/questions/ai-data-reach/) | AI workloads | 2 | Yes, access follows the user = 1; Scoped to the data store, not the user = 0.5; No or not known = 0 | high | | [27. Do destructive or high-impact agent actions (delete, grant access, pay, send externally) need a human approval?](/questions/human-approval/) | AI workloads | 2 | Yes, enforced in the tool = 1; Documented, not enforced = 0.5; No or not known = 0 | medium | | [28. Are model calls, tool calls and the identity behind each logged and kept?](/questions/ai-activity-logging/) | AI workloads | 2 | Yes, logged and retained = 1; Partly logged = 0.5; No or not known = 0 | low | | [29. Do organization-level policies block the riskiest actions before they happen (SCPs, Azure Policy deny effects, organization policy constraints)?](/questions/preventive-policies/) | Guardrails and drift | 3 | Yes, deny rules cover our top risks = 1; A few deny rules = 0.5; No or not known = 0 | medium | | [30. Do infrastructure changes go through code review and pipelines, with console changes in production flagged?](/questions/changes-as-code/) | Guardrails and drift | 2 | Yes, console writes are flagged = 1; Mostly code, console not monitored = 0.5; No or not known = 0 | high | | [31. Does a continuous check compare live configuration against your rules and report differences within a day?](/questions/drift-detection/) | Guardrails and drift | 2 | Yes, continuous with owners = 1; Periodic scans = 0.5; No or not known = 0 | medium | | [32. Do you test a new guardrail in audit or report-only mode and see what it would block before you enforce it?](/questions/guardrail-testing/) | Guardrails and drift | 1 | Yes, always = 1; Sometimes = 0.5; No or not known = 0 | medium | | [33. Does every guardrail exception have an owner, a reason and an expiry date?](/questions/exceptions-expire/) | Guardrails and drift | 1 | Yes, tracked with expiry = 1; Tracked, no expiry = 0.5; No or not known = 0 | low | | [34. Can administrators inside a workload account not turn off logging, guardrails or security services?](/questions/tamper-protection/) | Guardrails and drift | 2 | Yes, protected from account admins = 1; Protected for some controls = 0.5; No or not known = 0 | medium | | [35. Are control-plane audit logs on in every account, subscription and project, in every region, including data access logs for sensitive stores?](/questions/audit-logs-everywhere/) | Evidence | 3 | Yes, everywhere = 1; In most places = 0.5; No or not known = 0 | low | | [36. Are audit logs stored in a separate, restricted account where workload administrators cannot change or delete them?](/questions/logs-protected/) | Evidence | 2 | Yes, separate and write-once = 1; Separate, but not write-once = 0.5; No or not known = 0 | medium | | [37. Do you keep audit logs for at least 365 days, with the most recent 90 days searchable?](/questions/log-retention/) | Evidence | 2 | Yes, 365 days or more = 1; Under 365 days = 0.5; No or not known = 0 | low | | [38. Do alerts fire, and reach a person, for root or global administrator sign-ins, policy changes, new keys and disabled logging?](/questions/identity-alerts/) | Evidence | 2 | Yes, all four route to on-call = 1; Some are alerted = 0.5; No or not known = 0 | medium | | [39. Within one working day, can you produce a list of who can access a given sensitive data store and how they got that access?](/questions/access-evidence-on-demand/) | Evidence | 1 | Yes, within a day = 1; Within a few weeks = 0.5; No or not known = 0 | medium | | [40. Have you practiced cutting off all access of one compromised identity within one hour, and written down the steps?](/questions/revocation-drill/) | Evidence | 1 | Yes, in the past 12 months = 1; Planned, not practiced = 0.5; No or not known = 0 | high | ## Twenty known cases The scoring is checked against twenty cases kept in the site's tests folder, written by hand from the model above; the build fails if one gives a different answer. Cite this line: 'Cloud Security Risk Assessment, the 40-question assessment, edition 16 Nov 2026, https://cloudsecurityriskassessment.com/methodology/' ## Related pages - [Cloud access risk register template (XLSX)](/risk-register/) - [30 questions about cloud access risk assessment](/faq/) - [The 40-question assessment: the data, downloadable](/datasets/) ## Sources - CIS AWS Foundations Benchmark v5.0 1.4 and 1.5: https://www.cisecurity.org/benchmark/amazon_web_services (read 16 Nov 2026) - Microsoft cloud security benchmark IM-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - Google Cloud security best practices Service account keys: https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) - Microsoft cloud security benchmark IM-3: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - Microsoft cloud security benchmark PA-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - Microsoft cloud security benchmark PA-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - Microsoft cloud security benchmark PA-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - Microsoft cloud security benchmark PA-5: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/methodology/ --- # The 40 questions, each with the control it comes from Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The 40 questions of the assessment, grouped by domain. Each question page gives the question, why it matters, the three answers, the weight, the control it comes from, the fix with its effort, and a test for AWS, Azure and Google Cloud, so a team can verify an answer instead of guessing. ## Identities and entitlements - [Does the root user (or Global Administrator, or Organization Admin) of every account have phishing-resistant MFA and no access keys?](/questions/root-account-mfa/): CIS AWS Foundations Benchmark v5.0 1.4 and 1.5 - [Do all people reach the cloud through central single sign-on with MFA, with no local IAM users or standalone cloud passwords?](/questions/workforce-sso-mfa/): Microsoft cloud security benchmark IM-6 - [Are long-lived access keys and service account keys absent, or rotated at least every 90 days and unused ones deleted?](/questions/long-lived-keys/): Google Cloud security best practices Service account keys - [Do workloads get cloud access through attached roles, managed identities or workload identity federation instead of stored secrets?](/questions/workload-identity/): Microsoft cloud security benchmark IM-3 - [Is the number of people with administrator-level roles small, named and separate from their daily accounts?](/questions/admin-roles-limited/): Microsoft cloud security benchmark PA-1 - [Are privileged roles granted only for a limited time, on request, instead of held permanently?](/questions/no-standing-privilege/): Microsoft cloud security benchmark PA-2 - [Are permissions reviewed at least every 90 days and unused ones removed?](/questions/access-reviews/): Microsoft cloud security benchmark PA-4 - [Are emergency (break-glass) accounts in place, protected, monitored and tested at least twice a year?](/questions/emergency-access/): Microsoft cloud security benchmark PA-5 ## Network reachability - [Is every SSH, RDP and database port closed to the whole internet (0.0.0.0/0 and ::/0)?](/questions/no-open-admin-ports/): Google Cloud security best practices VPC firewall rules - [Do your data stores and platform services (storage, databases, secrets, AI) accept traffic only through private endpoints?](/questions/private-service-endpoints/): Microsoft cloud security benchmark NS-2 - [Do new compute instances and databases get no public IP address by default?](/questions/no-default-public-ip/): Microsoft cloud security benchmark NS-1 - [Do administrators reach servers through a brokered, logged service (Session Manager, Bastion, Identity-Aware Proxy) with no inbound admin ports?](/questions/brokered-admin-access/): Microsoft cloud security benchmark PA-6 - [Is the instance metadata service hardened (IMDSv2 required on AWS, managed identity endpoint protected on Azure, metadata concealment on GKE)?](/questions/metadata-service-hardened/): AWS documentation Instance metadata - [Is outbound traffic from production workloads filtered to known destinations?](/questions/egress-controlled/): Microsoft cloud security benchmark NS-3 - [Do you analyze which resources are reachable from the internet and from other networks (peering, transit, cross-account) at least monthly?](/questions/reachability-analysis/): Microsoft cloud security benchmark NS-7 ## Data perimeter - [Is public access to object storage blocked at the organization level, not just per bucket?](/questions/public-storage-blocked/): AWS documentation S3 Block Public Access - [Do your resource policies and sharing rules limit access to identities in your own organization?](/questions/organization-boundary/): AWS documentation Data perimeter - [Can your most sensitive data be read only from trusted networks or endpoints (VPC endpoints, service perimeters, private links)?](/questions/trusted-network-condition/): Google Cloud security best practices VPC Service Controls - [Do you review findings about resources shared outside your organization at least weekly?](/questions/external-access-findings/): AWS documentation IAM Access Analyzer - [Are sensitive data encrypted with customer-managed keys whose use is limited to named identities and services?](/questions/key-control/): Microsoft cloud security benchmark DP-5 - [Do you know which storage locations hold regulated or sensitive data, and is that list up to date?](/questions/sensitive-data-located/): Microsoft cloud security benchmark DP-1 - [Is every third-party and cross-account role protected by an external ID or condition, with an owner and an end date?](/questions/third-party-access/): AWS documentation IAM best practices ## AI workloads - [Does each AI agent, pipeline and application have its own cloud identity, not a shared one or a person's credentials?](/questions/ai-own-identity/): Microsoft cloud security benchmark AI-4 - [Is the list of actions each AI agent can perform limited to what its task needs, and written down?](/questions/ai-tool-permissions/): Microsoft cloud security benchmark AI-4 - [Can only approved models and AI services be used in your accounts, enforced by policy?](/questions/approved-models-only/): Microsoft cloud security benchmark AI-1 - [Is the data an AI system can read (retrieval stores, training data, tool outputs) limited to what its users may see?](/questions/ai-data-reach/): Google Cloud security best practices AI and ML security - [Do destructive or high-impact agent actions (delete, grant access, pay, send externally) need a human approval?](/questions/human-approval/): Microsoft cloud security benchmark AI-5 - [Are model calls, tool calls and the identity behind each logged and kept?](/questions/ai-activity-logging/): Microsoft cloud security benchmark AI-6 ## Guardrails and drift - [Do organization-level policies block the riskiest actions before they happen (SCPs, Azure Policy deny effects, organization policy constraints)?](/questions/preventive-policies/): AWS documentation Service control policies - [Do infrastructure changes go through code review and pipelines, with console changes in production flagged?](/questions/changes-as-code/): Microsoft cloud security benchmark PV-1 - [Does a continuous check compare live configuration against your rules and report differences within a day?](/questions/drift-detection/): AWS documentation AWS Config - [Do you test a new guardrail in audit or report-only mode and see what it would block before you enforce it?](/questions/guardrail-testing/): Microsoft documentation Azure Policy - [Does every guardrail exception have an owner, a reason and an expiry date?](/questions/exceptions-expire/): Microsoft cloud security benchmark PV-2 - [Can administrators inside a workload account not turn off logging, guardrails or security services?](/questions/tamper-protection/): Microsoft cloud security benchmark PV-2 ## Evidence - [Are control-plane audit logs on in every account, subscription and project, in every region, including data access logs for sensitive stores?](/questions/audit-logs-everywhere/): Microsoft cloud security benchmark LT-3 - [Are audit logs stored in a separate, restricted account where workload administrators cannot change or delete them?](/questions/logs-protected/): Microsoft cloud security benchmark LT-5 - [Do you keep audit logs for at least 365 days, with the most recent 90 days searchable?](/questions/log-retention/): Microsoft cloud security benchmark LT-6 - [Do alerts fire, and reach a person, for root or global administrator sign-ins, policy changes, new keys and disabled logging?](/questions/identity-alerts/): Microsoft cloud security benchmark LT-2 - [Within one working day, can you produce a list of who can access a given sensitive data store and how they got that access?](/questions/access-evidence-on-demand/): Microsoft cloud security benchmark PA-4 - [Have you practiced cutting off all access of one compromised identity within one hour, and written down the steps?](/questions/revocation-drill/): NIST SP 800-53 Rev. 5 IR-4 ## Related pages - [Cloud security assessment tools and options](/options/) - [Guides](/guides/) - [Glossary](/glossary/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/ --- # Does the root user (or Global Administrator, or Organization Admin) of every account have phishing-resistant MFA and no access keys? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 1 of the assessment asks: Does the root user (or Global Administrator, or Organization Admin) of every account have phishing-resistant MFA and no access keys? The root or top administrator identity cannot be limited by normal policies, so one phished password or leaked key hands over the whole account. ## Why it matters The root or top administrator identity cannot be limited by normal policies, so one phished password or leaked key hands over the whole account. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, in every account | 1 | | In some accounts | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** In the IAM console credential report, check that the root row shows mfa_active true and access_key_1_active and access_key_2_active false, in every account of the organization. - **Azure:** In Entra ID, list users holding Global Administrator and check each has a phishing-resistant authentication strength (FIDO2 or certificate) in Conditional Access. - **Google Cloud:** In the Cloud Identity or Workspace admin console, check that every super administrator has security keys enforced under 2-step verification, and that no super administrator is used for daily work. ## The fix Register a hardware or passkey MFA device on every root or top administrator identity and delete any root access keys. Effort: low. ## Where it comes from - CIS AWS Foundations Benchmark v5.0 1.4 and 1.5: [No root access keys and MFA on root](https://www.cisecurity.org/benchmark/amazon_web_services) - NIST SP 800-53 Rev. 5 IA-2(1): [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 2: Identities and entitlements, workforce sso mfa](/questions/workforce-sso-mfa/) - [Question 3: Identities and entitlements, long lived keys](/questions/long-lived-keys/) - [Question 4: Identities and entitlements, workload identity](/questions/workload-identity/) ## Sources - CIS AWS Foundations Benchmark v5.0 1.4 and 1.5: https://www.cisecurity.org/benchmark/amazon_web_services (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 IA-2(1): https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/root-account-mfa/ --- # Do all people reach the cloud through central single sign-on with MFA, with no local IAM users or standalone cloud passwords? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 2 of the assessment asks: Do all people reach the cloud through central single sign-on with MFA, with no local IAM users or standalone cloud passwords? Local users sit outside your joiner and leaver process and outside your MFA rules, so they are where forgotten access builds up. ## Why it matters Local users sit outside your joiner and leaver process and outside your MFA rules, so they are where forgotten access builds up. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, all people use SSO with MFA | 1 | | Mostly, with some local users | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List IAM users (aws iam list-users) and IAM Identity Center assignments, and check that no IAM user with a console password exists outside break-glass accounts. - **Azure:** In Entra ID, filter users by type and sign-in method, and check that Conditional Access requires MFA for all users and all cloud apps. - **Google Cloud:** In Cloud Identity, check that Google Cloud sign-in is bound to your SSO profile, and that IAM policies contain no consumer accounts (gmail.com). ## The fix Federate every cloud account to the central identity provider with MFA and delete or disable all local human users. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark IM-6: [Use strong authentication controls](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management) - NIST SP 800-53 Rev. 5 IA-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 3: Identities and entitlements, long lived keys](/questions/long-lived-keys/) - [Question 4: Identities and entitlements, workload identity](/questions/workload-identity/) - [Question 5: Identities and entitlements, admin roles limit](/questions/admin-roles-limited/) ## Sources - Microsoft cloud security benchmark IM-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 IA-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/workforce-sso-mfa/ --- # Are long-lived access keys and service account keys absent, or rotated at least every 90 days and unused ones deleted? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 3 of the assessment asks: Are long-lived access keys and service account keys absent, or rotated at least every 90 days and unused ones deleted? A static key works from anywhere, never expires on its own and often ends up in code, tickets and laptops. ## Why it matters A static key works from anywhere, never expires on its own and often ends up in code, tickets and laptops. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, none or all under 90 days | 1 | | Some are older or unused | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Download the IAM credential report and sort by access_key_last_rotated and access_key_last_used; every active key older than 90 days is a finding. - **Azure:** List app registration client secrets and certificates with their expiry dates (Entra admin center, App registrations), and flag any valid for more than 12 months. - **Google Cloud:** Run gcloud iam service-accounts keys list for each service account with --managed-by=user, and use the Policy Intelligence service account insights to find unused keys. ## The fix Replace static keys with short-lived credentials, and delete every key older than 90 days or unused for 90 days. Effort: medium. ## Where it comes from - Google Cloud security best practices Service account keys: [Best practices for managing service account keys](https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys) - Microsoft cloud security benchmark IM-8: [source](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management) - NIST SP 800-53 Rev. 5 IA-5: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 4: Identities and entitlements, workload identity](/questions/workload-identity/) - [Question 5: Identities and entitlements, admin roles limit](/questions/admin-roles-limited/) - [Question 6: Identities and entitlements, no standing privi](/questions/no-standing-privilege/) ## Sources - Google Cloud security best practices Service account keys: https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) - Microsoft cloud security benchmark IM-8: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 IA-5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/long-lived-keys/ --- # Do workloads get cloud access through attached roles, managed identities or workload identity federation instead of stored secrets? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 4 of the assessment asks: Do workloads get cloud access through attached roles, managed identities or workload identity federation instead of stored secrets? A workload that receives short-lived credentials from the platform has nothing to leak from a repository or a build log. ## Why it matters A workload that receives short-lived credentials from the platform has nothing to leak from a repository or a build log. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for all workloads | 1 | | For some workloads | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that EC2, ECS, EKS and Lambda use execution or instance roles and that CI systems use OIDC federation to assume roles instead of stored access keys. - **Azure:** Check that apps use system-assigned or user-assigned managed identities or federated credentials on app registrations instead of client secrets. - **Google Cloud:** Check that workloads use attached service accounts or workload identity federation, and that the iam.disableServiceAccountKeyCreation org policy is enforced. ## The fix Move each workload to platform-issued identities and remove the stored secret it used before. Effort: high. ## Where it comes from - Microsoft cloud security benchmark IM-3: [Manage application identities securely and automatically](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management) - NIST SP 800-53 Rev. 5 IA-5: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 5: Identities and entitlements, admin roles limit](/questions/admin-roles-limited/) - [Question 6: Identities and entitlements, no standing privi](/questions/no-standing-privilege/) - [Question 7: Identities and entitlements, access reviews](/questions/access-reviews/) ## Sources - Microsoft cloud security benchmark IM-3: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 IA-5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/workload-identity/ --- # Is the number of people with administrator-level roles small, named and separate from their daily accounts? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 5 of the assessment asks: Is the number of people with administrator-level roles small, named and separate from their daily accounts? Every extra administrator widens the set of accounts whose compromise ends in full control. It comes from Microsoft cloud security benchmark PA-1 and carries weight 2 in the identities and entitlements domain. ## Why it matters Every extra administrator widens the set of accounts whose compromise ends in full control. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, small, named and separate | 1 | | Small but not separate | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List principals with AdministratorAccess or iam:* policies (IAM Access Analyzer or an IAM policy simulator export) and count them against your approved list. - **Azure:** In Entra ID, count active assignments of Owner, User Access Administrator and Global Administrator at root and management group scope. - **Google Cloud:** Run gcloud projects get-iam-policy and the organization equivalent, and count members with roles/owner or roles/editor. ## The fix Cut administrator assignments to a named list and give each administrator a separate account for daily work. Effort: low. ## Where it comes from - Microsoft cloud security benchmark PA-1: [Separate and limit highly privileged/administrative users](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-6: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 6: Identities and entitlements, no standing privi](/questions/no-standing-privilege/) - [Question 7: Identities and entitlements, access reviews](/questions/access-reviews/) - [Question 8: Identities and entitlements, emergency access](/questions/emergency-access/) ## Sources - Microsoft cloud security benchmark PA-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-6: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/admin-roles-limited/ --- # Are privileged roles granted only for a limited time, on request, instead of held permanently? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 6 of the assessment asks: Are privileged roles granted only for a limited time, on request, instead of held permanently? Standing privilege is available to an attacker at any hour, while time-limited access exists only when someone needs it. ## Why it matters Standing privilege is available to an attacker at any hour, while time-limited access exists only when someone needs it. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for all privileged roles | 1 | | For some privileged roles | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** In IAM Identity Center, check whether administrator permission sets are assigned permanently, or whether a request and approval workflow with a session duration limit is in place. - **Azure:** In Entra Privileged Identity Management, check that privileged roles show Eligible assignments with a maximum activation duration, not Active permanent. - **Google Cloud:** Check for Privileged Access Manager entitlements with a maximum grant duration, and for IAM bindings with time-based conditions on privileged roles. ## The fix Make privileged roles eligible only, with an approval and a maximum duration of a few hours. Effort: high. ## Where it comes from - Microsoft cloud security benchmark PA-2: [Avoid standing access for user accounts and permissions](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-6: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 7: Identities and entitlements, access reviews](/questions/access-reviews/) - [Question 8: Identities and entitlements, emergency access](/questions/emergency-access/) - [Question 9: Network reachability, no open admin ports](/questions/no-open-admin-ports/) ## Sources - Microsoft cloud security benchmark PA-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-6: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/no-standing-privilege/ --- # Are permissions reviewed at least every 90 days and unused ones removed? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 7 of the assessment asks: Are permissions reviewed at least every 90 days and unused ones removed? Access that nobody has used in months is risk with no benefit, and it is rarely removed unless a review forces it. ## Why it matters Access that nobody has used in months is risk with no benefit, and it is rarely removed unless a review forces it. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, reviewed and removed | 1 | | Reviewed, but removal is not enforced | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Use IAM Access Analyzer unused access findings and the last accessed data on roles and users to list permissions unused for 90 days. - **Azure:** Check that Entra access reviews exist for privileged roles and groups with auto-apply on completion. - **Google Cloud:** Use IAM recommender role recommendations and Policy Analyzer to list excess permissions, and check that someone acts on them. ## The fix Schedule a quarterly review of all privileged and cross-account access and remove what was unused. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark PA-4: [Review and reconcile user access regularly](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 8: Identities and entitlements, emergency access](/questions/emergency-access/) - [Question 9: Network reachability, no open admin ports](/questions/no-open-admin-ports/) - [Question 10: Network reachability, private service endpoin](/questions/private-service-endpoints/) ## Sources - Microsoft cloud security benchmark PA-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/access-reviews/ --- # Are emergency (break-glass) accounts in place, protected, monitored and tested at least twice a year? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 8 of the assessment asks: Are emergency (break-glass) accounts in place, protected, monitored and tested at least twice a year? If your identity provider fails or a policy locks everyone out, a tested emergency account is the only way back in. ## Why it matters If your identity provider fails or a policy locks everyone out, a tested emergency account is the only way back in. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, protected, alerted and tested | 1 | | Exist but untested or unmonitored | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that the break-glass IAM users or roles exist, that a CloudWatch or EventBridge alert fires on their use, and note the last test date. - **Azure:** Check that two cloud-only emergency accounts are excluded from Conditional Access policies by design and have a sign-in alert in Azure Monitor. - **Google Cloud:** Check for a documented emergency super administrator in Cloud Identity with hardware keys stored offline, and a Cloud Logging alert on its use. ## The fix Create two emergency accounts with long unique credentials, alert on every sign-in and test them twice a year. Effort: low. ## Where it comes from - Microsoft cloud security benchmark PA-5: [Set up emergency access](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 9: Network reachability, no open admin ports](/questions/no-open-admin-ports/) - [Question 10: Network reachability, private service endpoin](/questions/private-service-endpoints/) - [Question 11: Network reachability, no default public ip](/questions/no-default-public-ip/) ## Sources - Microsoft cloud security benchmark PA-5: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/emergency-access/ --- # Is every SSH, RDP and database port closed to the whole internet (0.0.0.0/0 and ::/0)? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 9 of the assessment asks: Is every SSH, RDP and database port closed to the whole internet (0.0.0.0/0 and ::/0)? Open administrative ports are scanned within minutes of appearing, and a single weak or leaked credential is then enough. ## Why it matters Open administrative ports are scanned within minutes of appearing, and a single weak or leaked credential is then enough. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, none are open | 1 | | Open on a few exceptions | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Describe security groups with ip-permission.cidr 0.0.0.0/0 on ports 22, 3389, 3306, 5432 and 1433, or use the matching AWS Config managed rules. - **Azure:** In Network security groups, search inbound rules with source Any or Internet on those ports, or use the Defender for Cloud recommendation for management ports. - **Google Cloud:** List firewall rules (gcloud compute firewall-rules list) with source range 0.0.0.0/0 and allowed ports 22, 3389 or database ports. ## The fix Remove every inbound rule from 0.0.0.0/0 and ::/0 on ports 22, 3389 and database ports, and use brokered access instead. Effort: low. ## Where it comes from - Google Cloud security best practices VPC firewall rules: [VPC firewall rules overview](https://cloud.google.com/vpc/docs/firewalls) - Microsoft cloud security benchmark NS-1: [source](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security) - NIST SP 800-53 Rev. 5 SC-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 10: Network reachability, private service endpoin](/questions/private-service-endpoints/) - [Question 11: Network reachability, no default public ip](/questions/no-default-public-ip/) - [Question 12: Network reachability, brokered admin access](/questions/brokered-admin-access/) ## Sources - Google Cloud security best practices VPC firewall rules: https://cloud.google.com/vpc/docs/firewalls (read 16 Nov 2026) - Microsoft cloud security benchmark NS-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/no-open-admin-ports/ --- # Do your data stores and platform services (storage, databases, secrets, AI) accept traffic only through private endpoints? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 10 of the assessment asks: Do your data stores and platform services (storage, databases, secrets, AI) accept traffic only through private endpoints? A service with a public endpoint depends only on its credentials, while a private endpoint adds a network condition an attacker must also meet. ## Why it matters A service with a public endpoint depends only on its credentials, while a private endpoint adds a network condition an attacker must also meet. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for all production services | 1 | | For some services | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check for interface and gateway VPC endpoints for S3, DynamoDB, Secrets Manager and Bedrock, and for resource policies that deny access outside the VPC. - **Azure:** Check that Storage, SQL and Key Vault show Private endpoint connections and Public network access set to Disabled. - **Google Cloud:** Check for Private Service Connect endpoints and Private Google Access, and for VPC Service Controls perimeters around the services. ## The fix Create private endpoints for production data services and turn off their public network access. Effort: high. ## Where it comes from - Microsoft cloud security benchmark NS-2: [Secure cloud native services with network controls](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security) - NIST SP 800-53 Rev. 5 SC-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 11: Network reachability, no default public ip](/questions/no-default-public-ip/) - [Question 12: Network reachability, brokered admin access](/questions/brokered-admin-access/) - [Question 13: Network reachability, metadata service harden](/questions/metadata-service-hardened/) ## Sources - Microsoft cloud security benchmark NS-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/private-service-endpoints/ --- # Do new compute instances and databases get no public IP address by default? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 11 of the assessment asks: Do new compute instances and databases get no public IP address by default? If a public address needs an explicit decision, fewer resources are exposed by accident. It comes from Microsoft cloud security benchmark NS-1 and carries weight 2 in the network reachability domain. ## Why it matters If a public address needs an explicit decision, fewer resources are exposed by accident. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, enforced by policy | 1 | | By convention, not enforced | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List instances and network interfaces with a public IP (aws ec2 describe-instances) and check subnet settings for auto-assign public IP, or the matching AWS Config rules. - **Azure:** Query Azure Resource Graph for network interfaces with a public IP address attached and for databases with public access enabled. - **Google Cloud:** List instances with external IPs (gcloud compute instances list) and check that the compute.vmExternalIpAccess org policy constraint is enforced. ## The fix Enforce a policy that denies public IP assignment on compute and databases unless an approved exception exists. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark NS-1: [Establish network segmentation boundaries](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security) - NIST SP 800-53 Rev. 5 SC-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 12: Network reachability, brokered admin access](/questions/brokered-admin-access/) - [Question 13: Network reachability, metadata service harden](/questions/metadata-service-hardened/) - [Question 14: Network reachability, egress controlled](/questions/egress-controlled/) ## Sources - Microsoft cloud security benchmark NS-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/no-default-public-ip/ --- # Do administrators reach servers through a brokered, logged service (Session Manager, Bastion, Identity-Aware Proxy) with no inbound admin ports? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 12 of the assessment asks: Do administrators reach servers through a brokered, logged service (Session Manager, Bastion, Identity-Aware Proxy) with no inbound admin ports? A broker turns each login into an identity-checked, logged session instead of a network path open to anyone with a key. ## Why it matters A broker turns each login into an identity-checked, logged session instead of a network path open to anyone with a key. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for all servers | 1 | | For some servers | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that instances are managed by AWS Systems Manager Session Manager, and that session logging to S3 or CloudWatch is on. - **Azure:** Check that Azure Bastion or just-in-time VM access is used and that NSGs have no inbound admin rules. - **Google Cloud:** Check that Identity-Aware Proxy TCP forwarding or OS Login is used for VM access, with the matching IAM roles granted. ## The fix Switch server administration to a brokered service and delete shared SSH keys and jump hosts. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark PA-6: [Use privileged access solution](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-17: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 13: Network reachability, metadata service harden](/questions/metadata-service-hardened/) - [Question 14: Network reachability, egress controlled](/questions/egress-controlled/) - [Question 15: Network reachability, reachability analysis](/questions/reachability-analysis/) ## Sources - Microsoft cloud security benchmark PA-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-17: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/brokered-admin-access/ --- # Is the instance metadata service hardened (IMDSv2 required on AWS, managed identity endpoint protected on Azure, metadata concealment on GKE)? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 13 of the assessment asks: Is the instance metadata service hardened (IMDSv2 required on AWS, managed identity endpoint protected on Azure, metadata concealment on GKE)? A server-side request forgery bug can read the metadata service and walk away with the instance's cloud credentials. ## Why it matters A server-side request forgery bug can read the metadata service and walk away with the instance's cloud credentials. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, enforced everywhere | 1 | | On new resources only | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Describe instances and check MetadataOptions.HttpTokens is required, and that the account default under EC2 data protection and security is set to required. - **Azure:** Check that workloads do not accept unvalidated URLs, that managed identity use is scoped per resource, and that Defender for Cloud has no open server-side request findings. - **Google Cloud:** Check that GKE clusters use Workload Identity (metadata server mode GKE_METADATA) and that Compute Engine instances do not run with the default service account. ## The fix Require IMDSv2 on all instances and set the account default so new instances cannot opt out. Effort: low. ## Where it comes from - AWS documentation Instance metadata: [Configure the instance metadata service](https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configuring-instance-metadata-service.html) - Microsoft cloud security benchmark IM-8: [source](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management) ## Related pages - [Question 14: Network reachability, egress controlled](/questions/egress-controlled/) - [Question 15: Network reachability, reachability analysis](/questions/reachability-analysis/) - [Question 16: Data perimeter, public storage blocked](/questions/public-storage-blocked/) ## Sources - AWS documentation Instance metadata: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configuring-instance-metadata-service.html (read 16 Nov 2026) - Microsoft cloud security benchmark IM-8: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/metadata-service-hardened/ --- # Is outbound traffic from production workloads filtered to known destinations? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 14 of the assessment asks: Is outbound traffic from production workloads filtered to known destinations? Egress control limits what a compromised workload can send out and which attacker servers it can call. It comes from Microsoft cloud security benchmark NS-3 and carries weight 1 in the network reachability domain. ## Why it matters Egress control limits what a compromised workload can send out and which attacker servers it can call. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, deny by default with an allow list | 1 | | Partly, for some networks | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check route tables for NAT gateways and AWS Network Firewall or Route 53 Resolver DNS Firewall rules that restrict outbound destinations. - **Azure:** Check for Azure Firewall or a network virtual appliance with user-defined routes sending 0.0.0.0/0 through it. - **Google Cloud:** Check for Cloud NAT with Cloud Firewall egress deny rules and fully qualified domain name allow lists on the production VPC. ## The fix Route production egress through a firewall or proxy with a destination allow list. Effort: high. ## Where it comes from - Microsoft cloud security benchmark NS-3: [Deploy firewall at the edge of enterprise network](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security) - NIST SP 800-53 Rev. 5 SC-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 15: Network reachability, reachability analysis](/questions/reachability-analysis/) - [Question 16: Data perimeter, public storage blocked](/questions/public-storage-blocked/) - [Question 17: Data perimeter, organization boundary](/questions/organization-boundary/) ## Sources - Microsoft cloud security benchmark NS-3: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/egress-controlled/ --- # Do you analyze which resources are reachable from the internet and from other networks (peering, transit, cross-account) at least monthly? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 15 of the assessment asks: Do you analyze which resources are reachable from the internet and from other networks (peering, transit, cross-account) at least monthly? Individual rules look safe on their own, and reachability only shows up when you evaluate the paths together. ## Why it matters Individual rules look safe on their own, and reachability only shows up when you evaluate the paths together. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, on a schedule with owners | 1 | | Ad hoc | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Use VPC Reachability Analyzer and Network Access Analyzer scopes for internet-reachable and cross-VPC paths. - **Azure:** Use Azure Network Watcher connection troubleshoot and effective security rules, and Defender for Cloud attack path views where available. - **Google Cloud:** Use Network Intelligence Center Connectivity Tests and Firewall Insights for rule hit counts and shadowed rules. ## The fix Run a path-based reachability analysis monthly and assign every internet-reachable resource an owner. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark NS-7: [Manage network security centrally and effectively](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security) - NIST SP 800-53 Rev. 5 SC-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 16: Data perimeter, public storage blocked](/questions/public-storage-blocked/) - [Question 17: Data perimeter, organization boundary](/questions/organization-boundary/) - [Question 18: Data perimeter, trusted network condition](/questions/trusted-network-condition/) ## Sources - Microsoft cloud security benchmark NS-7: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-network-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/reachability-analysis/ --- # Is public access to object storage blocked at the organization level, not just per bucket? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 16 of the assessment asks: Is public access to object storage blocked at the organization level, not just per bucket? A single public bucket can expose an entire dataset, and a block at the top is the only one that applies to buckets created later. ## Why it matters A single public bucket can expose an entire dataset, and a block at the top is the only one that applies to buckets created later. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, organization-wide | 1 | | On most accounts or buckets | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check S3 Block Public Access at the account level in every account, or the organization-level policy, with all four settings on. - **Azure:** Check that every storage account has Allow Blob anonymous access set to Disabled, and that Azure Policy denies the setting. - **Google Cloud:** Check that the storage.publicAccessPrevention org policy constraint is enforced and that buckets show Public access prevention on. ## The fix Turn on the organization-level public access block for object storage and fix the buckets that break. Effort: low. ## Where it comes from - AWS documentation S3 Block Public Access: [Blocking public access to your Amazon S3 storage](https://docs.aws.amazon.com/AmazonS3/latest/userguide/access-control-block-public-access.html) - NIST SP 800-53 Rev. 5 AC-3: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 17: Data perimeter, organization boundary](/questions/organization-boundary/) - [Question 18: Data perimeter, trusted network condition](/questions/trusted-network-condition/) - [Question 19: Data perimeter, external access findings](/questions/external-access-findings/) ## Sources - AWS documentation S3 Block Public Access: https://docs.aws.amazon.com/AmazonS3/latest/userguide/access-control-block-public-access.html (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-3: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/public-storage-blocked/ --- # Do your resource policies and sharing rules limit access to identities in your own organization? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 17 of the assessment asks: Do your resource policies and sharing rules limit access to identities in your own organization? Without an organization boundary, one mistaken policy line can give any account on the internet access to your data. ## Why it matters Without an organization boundary, one mistaken policy line can give any account on the internet access to your data. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, enforced by policy | 1 | | In some policies only | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check SCPs or resource control policies for aws:PrincipalOrgID and aws:ResourceOrgID conditions, and test with Access Analyzer external access findings. - **Azure:** Check tenant restrictions, B2B cross-tenant access settings and Azure Policy on storage, and review which external tenants are allowed. - **Google Cloud:** Check that the iam.allowedPolicyMemberDomains org policy (domain restricted sharing) is enforced and lists only your own customer IDs. ## The fix Add an organization-boundary condition to resource policies and a deny rule at the organization level. Effort: medium. ## Where it comes from - AWS documentation Data perimeter: [Building a data perimeter on AWS](https://docs.aws.amazon.com/whitepapers/latest/building-a-data-perimeter-on-aws/building-a-data-perimeter-on-aws.html) - NIST SP 800-53 Rev. 5 AC-4: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 18: Data perimeter, trusted network condition](/questions/trusted-network-condition/) - [Question 19: Data perimeter, external access findings](/questions/external-access-findings/) - [Question 20: Data perimeter, key control](/questions/key-control/) ## Sources - AWS documentation Data perimeter: https://docs.aws.amazon.com/whitepapers/latest/building-a-data-perimeter-on-aws/building-a-data-perimeter-on-aws.html (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-4: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/organization-boundary/ --- # Can your most sensitive data be read only from trusted networks or endpoints (VPC endpoints, service perimeters, private links)? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 18 of the assessment asks: Can your most sensitive data be read only from trusted networks or endpoints (VPC endpoints, service perimeters, private links)? A stolen credential used from the internet fails when the data only answers inside your network boundary. ## Why it matters A stolen credential used from the internet fails when the data only answers inside your network boundary. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for all sensitive data | 1 | | For some data stores | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check VPC endpoint policies and bucket policies using aws:SourceVpce, aws:SourceVpc or aws:SourceIp conditions on sensitive buckets and keys. - **Azure:** Check storage and Key Vault firewalls set to selected networks, and Conditional Access policies with named locations. - **Google Cloud:** List VPC Service Controls perimeters and access levels, and check the sensitive projects are inside one in enforced mode, not dry run. ## The fix Add network conditions to the policies of the sensitive data stores so they answer only from approved networks. Effort: high. ## Where it comes from - Google Cloud security best practices VPC Service Controls: [VPC Service Controls overview](https://cloud.google.com/vpc-service-controls/docs/overview) - NIST SP 800-53 Rev. 5 AC-4: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 19: Data perimeter, external access findings](/questions/external-access-findings/) - [Question 20: Data perimeter, key control](/questions/key-control/) - [Question 21: Data perimeter, sensitive data located](/questions/sensitive-data-located/) ## Sources - Google Cloud security best practices VPC Service Controls: https://cloud.google.com/vpc-service-controls/docs/overview (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-4: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/trusted-network-condition/ --- # Do you review findings about resources shared outside your organization at least weekly? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 19 of the assessment asks: Do you review findings about resources shared outside your organization at least weekly? Sharing outside the account is sometimes intended, and a regular review separates the intended cases from the leftover ones. It comes from AWS documentation IAM Access Analyzer and carries weight 2 in the data perimeter domain. ## Why it matters Sharing outside the account is sometimes intended, and a regular review separates the intended cases from the leftover ones. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, weekly or continuous with owners | 1 | | Occasionally | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that an IAM Access Analyzer for the organization exists with external access type, and count active findings and their age. - **Azure:** Review Entra cross-tenant access settings, guest users and Azure resources shared through SAS tokens or public network access. - **Google Cloud:** Use Cloud Asset Inventory searches for IAM bindings to allUsers, allAuthenticatedUsers and principals outside your domain. ## The fix Enable external access analysis at the organization level and route new findings to an owner with a due date. Effort: low. ## Where it comes from - AWS documentation IAM Access Analyzer: [Getting started with Access Analyzer](https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-getting-started.html) - NIST SP 800-53 Rev. 5 AC-3: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 20: Data perimeter, key control](/questions/key-control/) - [Question 21: Data perimeter, sensitive data located](/questions/sensitive-data-located/) - [Question 22: Data perimeter, third party access](/questions/third-party-access/) ## Sources - AWS documentation IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/access-analyzer-getting-started.html (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-3: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/external-access-findings/ --- # Are sensitive data encrypted with customer-managed keys whose use is limited to named identities and services? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 20 of the assessment asks: Are sensitive data encrypted with customer-managed keys whose use is limited to named identities and services? Key policy is a second, separate lock, so access to a data store alone does not give readable data. ## Why it matters Key policy is a second, separate lock, so access to a data store alone does not give readable data. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, for sensitive data | 1 | | For some data | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List KMS keys and check key policies and grants for kms:Decrypt by principals, and that sensitive S3, EBS and RDS resources use customer-managed keys. - **Azure:** Check Key Vault or Managed HSM access policy or RBAC assignments and that sensitive storage uses customer-managed keys with purge protection. - **Google Cloud:** Check Cloud KMS key IAM bindings for roles/cloudkms.cryptoKeyDecrypter and that sensitive buckets and datasets use CMEK. ## The fix Move sensitive data stores to customer-managed keys and restrict key use to the services and roles that need it. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark DP-5: [Use customer-managed key option in data at rest encryption when required](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-data-protection) - NIST SP 800-53 Rev. 5 SC-12: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 21: Data perimeter, sensitive data located](/questions/sensitive-data-located/) - [Question 22: Data perimeter, third party access](/questions/third-party-access/) - [Question 23: AI workloads, ai own identity](/questions/ai-own-identity/) ## Sources - Microsoft cloud security benchmark DP-5: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-data-protection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 SC-12: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/key-control/ --- # Do you know which storage locations hold regulated or sensitive data, and is that list up to date? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 21 of the assessment asks: Do you know which storage locations hold regulated or sensitive data, and is that list up to date? You cannot put a stricter boundary around data you cannot find, so locating it comes before most other data controls. ## Why it matters You cannot put a stricter boundary around data you cannot find, so locating it comes before most other data controls. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, scanned and tagged | 1 | | From memory or partly | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check for Amazon Macie jobs or equivalent scans of S3 and a tag standard such as data-classification on buckets and databases. - **Azure:** Check Microsoft Purview data map scans or Defender for Storage sensitive data discovery for storage and databases. - **Google Cloud:** Check Sensitive Data Protection discovery scan configurations on BigQuery, Cloud Storage and Cloud SQL. ## The fix Run sensitive data discovery on all storage and tag the locations it finds. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark DP-1: [Discover, classify, and label sensitive data](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-data-protection) - NIST SP 800-53 Rev. 5 AC-4: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 22: Data perimeter, third party access](/questions/third-party-access/) - [Question 23: AI workloads, ai own identity](/questions/ai-own-identity/) - [Question 24: AI workloads, ai tool permissions](/questions/ai-tool-permissions/) ## Sources - Microsoft cloud security benchmark DP-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-data-protection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-4: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/sensitive-data-located/ --- # Is every third-party and cross-account role protected by an external ID or condition, with an owner and an end date? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 22 of the assessment asks: Is every third-party and cross-account role protected by an external ID or condition, with an owner and an end date? Third-party access is trust you extend outside your own controls, and an unconditioned role can be used by anyone who learns its name. ## Why it matters Third-party access is trust you extend outside your own controls, and an unconditioned role can be used by anyone who learns its name. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, all have conditions, owners and dates | 1 | | Some do | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List roles whose trust policy names an external account, and check for sts:ExternalId or aws:PrincipalOrgID conditions. - **Azure:** List guest users and enterprise applications with delegated permissions, and review Azure Lighthouse and partner delegated access. - **Google Cloud:** List IAM bindings for principals outside your domain, and review Workload Identity Federation providers and who may impersonate service accounts. ## The fix Add an external ID or condition, an owner and an expiry date to every third-party trust, and remove those without an owner. Effort: medium. ## Where it comes from - AWS documentation IAM best practices: [Security best practices in IAM](https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html) - NIST SP 800-53 Rev. 5 AC-3: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 23: AI workloads, ai own identity](/questions/ai-own-identity/) - [Question 24: AI workloads, ai tool permissions](/questions/ai-tool-permissions/) - [Question 25: AI workloads, approved models only](/questions/approved-models-only/) ## Sources - AWS documentation IAM best practices: https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-3: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/third-party-access/ --- # Does each AI agent, pipeline and application have its own cloud identity, not a shared one or a person's credentials? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 23 of the assessment asks: Does each AI agent, pipeline and application have its own cloud identity, not a shared one or a person's credentials? If an agent shares an identity, you cannot tell what it did and you cannot cut its access without breaking other things. ## Why it matters If an agent shares an identity, you cannot tell what it did and you cannot cut its access without breaking other things. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, one identity per agent or application | 1 | | Some share an identity | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that each Bedrock agent, Lambda or container that calls models has its own execution role, and that CloudTrail userIdentity shows distinct roles per agent. - **Azure:** Check that each AI app uses its own managed identity and that Azure OpenAI or AI Foundry role assignments are not made to a shared principal. - **Google Cloud:** Check that each Vertex AI agent or job runs as its own service account, not the default Compute Engine service account. ## The fix Give every AI agent and pipeline a dedicated identity and remove the shared roles and personal tokens it used. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark AI-4: [Apply least privilege for agent functions](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) - NIST SP 800-53 Rev. 5 IA-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 24: AI workloads, ai tool permissions](/questions/ai-tool-permissions/) - [Question 25: AI workloads, approved models only](/questions/approved-models-only/) - [Question 26: AI workloads, ai data reach](/questions/ai-data-reach/) ## Sources - Microsoft cloud security benchmark AI-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 IA-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/ai-own-identity/ --- # Is the list of actions each AI agent can perform limited to what its task needs, and written down? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 24 of the assessment asks: Is the list of actions each AI agent can perform limited to what its task needs, and written down? An agent that can call any API will use it when it is tricked or wrong, so the limit has to sit in the permissions, not in the prompt. ## Why it matters An agent that can call any API will use it when it is tricked or wrong, so the limit has to sit in the permissions, not in the prompt. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, scoped and reviewed | 1 | | Partly scoped | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Review the agent's role policies and the action groups of each Bedrock agent for wildcard actions or resources, using Access Analyzer policy validation. - **Azure:** Review role assignments of each AI app identity and AI Foundry connections for broad roles such as Contributor or Owner. - **Google Cloud:** Review IAM bindings of each Vertex AI agent's service account for primitive or broad roles, using the IAM recommender. ## The fix Replace broad roles on agents with policies naming only the actions, resources and data each agent needs. Effort: high. ## Where it comes from - Microsoft cloud security benchmark AI-4: [Apply least privilege for agent functions](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) - NIST SP 800-53 Rev. 5 AC-6: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 25: AI workloads, approved models only](/questions/approved-models-only/) - [Question 26: AI workloads, ai data reach](/questions/ai-data-reach/) - [Question 27: AI workloads, human approval](/questions/human-approval/) ## Sources - Microsoft cloud security benchmark AI-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-6: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/ai-tool-permissions/ --- # Can only approved models and AI services be used in your accounts, enforced by policy? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 25 of the assessment asks: Can only approved models and AI services be used in your accounts, enforced by policy? Unapproved models and endpoints send your data to places you have not assessed. It comes from Microsoft cloud security benchmark AI-1 and carries weight 2 in the ai workloads domain. ## Why it matters Unapproved models and endpoints send your data to places you have not assessed. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, enforced by policy | 1 | | By guidance only | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check SCPs or IAM policies that limit bedrock:InvokeModel to approved model ARNs, and that other AI services are denied where not needed. - **Azure:** Check Azure Policy assignments that restrict model deployments and AI service resource types to the approved set. - **Google Cloud:** Check the Vertex AI model garden and org policy constraint for restricting allowed models and services. ## The fix Enforce an allow list of models and AI services with organization policies. Effort: low. ## Where it comes from - Microsoft cloud security benchmark AI-1: [Ensure use of approved models](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) - NIST SP 800-53 Rev. 5 CM-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 26: AI workloads, ai data reach](/questions/ai-data-reach/) - [Question 27: AI workloads, human approval](/questions/human-approval/) - [Question 28: AI workloads, ai activity logging](/questions/ai-activity-logging/) ## Sources - Microsoft cloud security benchmark AI-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 CM-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/approved-models-only/ --- # Is the data an AI system can read (retrieval stores, training data, tool outputs) limited to what its users may see? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 26 of the assessment asks: Is the data an AI system can read (retrieval stores, training data, tool outputs) limited to what its users may see? A retrieval system that reads with its own broad access can show a user data the user was never allowed to open. ## Why it matters A retrieval system that reads with its own broad access can show a user data the user was never allowed to open. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, access follows the user | 1 | | Scoped to the data store, not the user | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check knowledge base data sources, vector stores and S3 buckets used by Bedrock for the roles that read them, and whether metadata filtering enforces user permissions. - **Azure:** Check AI Search indexes and data sources for security trimming by user or group, and which identity reads the source storage. - **Google Cloud:** Check Vertex AI Search and RAG corpora for data store access controls, and which service account reads the underlying buckets. ## The fix Make retrieval and tool calls check the end user's permissions, and split sensitive data into separate stores. Effort: high. ## Where it comes from - Google Cloud security best practices AI and ML security: [Google Cloud AI and ML perspective: security](https://cloud.google.com/architecture/framework/perspectives/ai-ml/security) - Microsoft cloud security benchmark AI-4: [source](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) ## Related pages - [Question 27: AI workloads, human approval](/questions/human-approval/) - [Question 28: AI workloads, ai activity logging](/questions/ai-activity-logging/) - [Question 29: Guardrails and drift, preventive policies](/questions/preventive-policies/) ## Sources - Google Cloud security best practices AI and ML security: https://cloud.google.com/architecture/framework/perspectives/ai-ml/security (read 16 Nov 2026) - Microsoft cloud security benchmark AI-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/ai-data-reach/ --- # Do destructive or high-impact agent actions (delete, grant access, pay, send externally) need a human approval? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 27 of the assessment asks: Do destructive or high-impact agent actions (delete, grant access, pay, send externally) need a human approval? Approval is the last control when an agent's reasoning or inputs are wrong. It comes from Microsoft cloud security benchmark AI-5 and carries weight 2 in the ai workloads domain. ## Why it matters Approval is the last control when an agent's reasoning or inputs are wrong. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, enforced in the tool | 1 | | Documented, not enforced | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check agent action groups for user confirmation settings and that Step Functions or Lambda workflows include a manual approval for high-impact calls. - **Azure:** Check that agent tools call approval workflows (Logic Apps approvals or Power Automate) before high-impact actions. - **Google Cloud:** Check that agent tools call approval workflows or Cloud Workflows callbacks before high-impact actions. ## The fix Require an approval step in the tool or workflow for each destructive or high-impact action. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark AI-5: [Ensure human-in-the-loop](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) - NIST SP 800-53 Rev. 5 AC-6: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 28: AI workloads, ai activity logging](/questions/ai-activity-logging/) - [Question 29: Guardrails and drift, preventive policies](/questions/preventive-policies/) - [Question 30: Guardrails and drift, changes as code](/questions/changes-as-code/) ## Sources - Microsoft cloud security benchmark AI-5: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-6: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/human-approval/ --- # Are model calls, tool calls and the identity behind each logged and kept? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 28 of the assessment asks: Are model calls, tool calls and the identity behind each logged and kept? When an agent acts, the log is the only record of what it was asked and what it did. It comes from Microsoft cloud security benchmark AI-6 and carries weight 2 in the ai workloads domain. ## Why it matters When an agent acts, the log is the only record of what it was asked and what it did. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, logged and retained | 1 | | Partly logged | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check Bedrock model invocation logging to S3 or CloudWatch Logs, and CloudTrail data events for agent and model APIs. - **Azure:** Check diagnostic settings on Azure OpenAI and AI Foundry resources send request and audit logs to a Log Analytics workspace. - **Google Cloud:** Check Vertex AI request-response logging and that Data Access audit logs are on for the Vertex AI API. ## The fix Turn on invocation and tool-call logging for every AI service and keep it with your other audit logs. Effort: low. ## Where it comes from - Microsoft cloud security benchmark AI-6: [Establish monitoring and detection](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security) - NIST SP 800-53 Rev. 5 AU-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 29: Guardrails and drift, preventive policies](/questions/preventive-policies/) - [Question 30: Guardrails and drift, changes as code](/questions/changes-as-code/) - [Question 31: Guardrails and drift, drift detection](/questions/drift-detection/) ## Sources - Microsoft cloud security benchmark AI-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/ai-activity-logging/ --- # Do organization-level policies block the riskiest actions before they happen (SCPs, Azure Policy deny effects, organization policy constraints)? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 29 of the assessment asks: Do organization-level policies block the riskiest actions before they happen (SCPs, Azure Policy deny effects, organization policy constraints)? Detection tells you after a change, while a deny rule stops it, and the gap between the two is where incidents happen. ## Why it matters Detection tells you after a change, while a deny rule stops it, and the gap between the two is where incidents happen. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, deny rules cover our top risks | 1 | | A few deny rules | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List SCPs attached to the organization root and OUs (aws organizations list-policies) and check they deny actions such as disabling logging or leaving the organization. - **Azure:** List Azure Policy assignments with the Deny effect at management group scope. - **Google Cloud:** List organization policy constraints enforced at the organization node (gcloud org-policies list) and check the security-relevant ones are set. ## The fix Write deny policies for your five riskiest actions at the organization root and apply them to every account. Effort: medium. ## Where it comes from - AWS documentation Service control policies: [Service control policies (SCPs)](https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html) - Microsoft cloud security benchmark PV-2: [source](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management) ## Related pages - [Question 30: Guardrails and drift, changes as code](/questions/changes-as-code/) - [Question 31: Guardrails and drift, drift detection](/questions/drift-detection/) - [Question 32: Guardrails and drift, guardrail testing](/questions/guardrail-testing/) ## Sources - AWS documentation Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) - Microsoft cloud security benchmark PV-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/preventive-policies/ --- # Do infrastructure changes go through code review and pipelines, with console changes in production flagged? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 30 of the assessment asks: Do infrastructure changes go through code review and pipelines, with console changes in production flagged? A change that bypasses review is a change nobody checked against your own rules. It comes from Microsoft cloud security benchmark PV-1 and carries weight 2 in the guardrails and drift domain. ## Why it matters A change that bypasses review is a change nobody checked against your own rules. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, console writes are flagged | 1 | | Mostly code, console not monitored | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Query CloudTrail for write events where userAgent or sessionContext indicates console or user sessions in production accounts. - **Azure:** Query the Activity Log for write operations by users rather than the pipeline service principal in production subscriptions. - **Google Cloud:** Query Cloud Audit Logs admin activity for write calls by user principals rather than the pipeline service account. ## The fix Alert on write actions from the console in production and make the pipeline the only role that can change infrastructure. Effort: high. ## Where it comes from - Microsoft cloud security benchmark PV-1: [Define and establish secure configurations](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management) - NIST SP 800-53 Rev. 5 CM-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 31: Guardrails and drift, drift detection](/questions/drift-detection/) - [Question 32: Guardrails and drift, guardrail testing](/questions/guardrail-testing/) - [Question 33: Guardrails and drift, exceptions expire](/questions/exceptions-expire/) ## Sources - Microsoft cloud security benchmark PV-1: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 CM-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/changes-as-code/ --- # Does a continuous check compare live configuration against your rules and report differences within a day? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 31 of the assessment asks: Does a continuous check compare live configuration against your rules and report differences within a day? Configuration drifts through emergency fixes and manual edits, and a check that runs once a year finds it a year late. ## Why it matters Configuration drifts through emergency fixes and manual edits, and a check that runs once a year finds it a year late. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, continuous with owners | 1 | | Periodic scans | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check AWS Config organization conformance packs or rules, with an aggregator across accounts and a record of non-compliant resources. - **Azure:** Check Azure Policy compliance state across management groups and that non-compliant resources are routed to owners. - **Google Cloud:** Check Security Command Center findings, Cloud Asset Inventory feeds or Organization Policy Analyzer for violations across the organization. ## The fix Run a continuous compliance check against your policy set and send non-compliant results to the resource owner. Effort: medium. ## Where it comes from - AWS documentation AWS Config: [What is AWS Config](https://docs.aws.amazon.com/config/latest/developerguide/WhatIsConfig.html) - NIST SP 800-53 Rev. 5 CA-7: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 32: Guardrails and drift, guardrail testing](/questions/guardrail-testing/) - [Question 33: Guardrails and drift, exceptions expire](/questions/exceptions-expire/) - [Question 34: Guardrails and drift, tamper protection](/questions/tamper-protection/) ## Sources - AWS documentation AWS Config: https://docs.aws.amazon.com/config/latest/developerguide/WhatIsConfig.html (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 CA-7: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/drift-detection/ --- # Do you test a new guardrail in audit or report-only mode and see what it would block before you enforce it? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 32 of the assessment asks: Do you test a new guardrail in audit or report-only mode and see what it would block before you enforce it? A guardrail enforced blind can stop production, and then it gets switched off for good. ## Why it matters A guardrail enforced blind can stop production, and then it gets switched off for good. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, always | 1 | | Sometimes | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check that SCP changes are tested in a sandbox OU, and use CloudTrail Lake or access last used data to see which calls a deny would have affected. - **Azure:** Check that new policy assignments start with the Audit effect or enforcement mode DoNotEnforce, and review compliance results first. - **Google Cloud:** Check that new organization policies are tested with dry run mode where available and with Policy Simulator before enforcement. ## The fix Run every new deny rule in audit mode for at least one week and review what it would have blocked. Effort: medium. ## Where it comes from - Microsoft documentation Azure Policy: [Overview of Azure Policy](https://learn.microsoft.com/en-us/azure/governance/policy/overview) - NIST SP 800-53 Rev. 5 CM-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 33: Guardrails and drift, exceptions expire](/questions/exceptions-expire/) - [Question 34: Guardrails and drift, tamper protection](/questions/tamper-protection/) - [Question 35: Evidence, audit logs everywhere](/questions/audit-logs-everywhere/) ## Sources - Microsoft documentation Azure Policy: https://learn.microsoft.com/en-us/azure/governance/policy/overview (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 CM-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/guardrail-testing/ --- # Does every guardrail exception have an owner, a reason and an expiry date? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 33 of the assessment asks: Does every guardrail exception have an owner, a reason and an expiry date? Exceptions without an end date become the permanent state. It comes from Microsoft cloud security benchmark PV-2 and carries weight 1 in the guardrails and drift domain. ## Why it matters Exceptions without an end date become the permanent state. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, tracked with expiry | 1 | | Tracked, no expiry | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** List SCP exemption conditions and AWS Config rule exclusions and check each against a register with an owner and an end date. - **Azure:** List Azure Policy exemptions (az policy exemption list) and check each has an expiry and a category. - **Google Cloud:** List organization policy overrides at folder and project level and check each against a register with an end date. ## The fix Record each exception with an owner, a reason and an end date, and remove it automatically at the end date. Effort: low. ## Where it comes from - Microsoft cloud security benchmark PV-2: [Audit and enforce secure configurations](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management) - NIST SP 800-53 Rev. 5 CM-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 34: Guardrails and drift, tamper protection](/questions/tamper-protection/) - [Question 35: Evidence, audit logs everywhere](/questions/audit-logs-everywhere/) - [Question 36: Evidence, logs protected](/questions/logs-protected/) ## Sources - Microsoft cloud security benchmark PV-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 CM-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/exceptions-expire/ --- # Can administrators inside a workload account not turn off logging, guardrails or security services? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 34 of the assessment asks: Can administrators inside a workload account not turn off logging, guardrails or security services? A control the account owner can disable protects against mistakes, but not against a compromised account owner. It comes from Microsoft cloud security benchmark PV-2 and carries weight 2 in the guardrails and drift domain. ## Why it matters A control the account owner can disable protects against mistakes, but not against a compromised account owner. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, protected from account admins | 1 | | Protected for some controls | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check SCPs deny cloudtrail:StopLogging, config:StopConfigurationRecorder and guardduty:DeleteDetector except for a security role. - **Azure:** Check resource locks and Azure Policy that prevent deletion of diagnostic settings, and who holds Policy Contributor. - **Google Cloud:** Check organization policy and deny policies preventing sink deletion and log bucket changes, plus the folder-level IAM for the security team only. ## The fix Deny changes to logging, security services and guardrails from every identity except a dedicated security role. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark PV-2: [Audit and enforce secure configurations](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management) - NIST SP 800-53 Rev. 5 AU-9: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 35: Evidence, audit logs everywhere](/questions/audit-logs-everywhere/) - [Question 36: Evidence, logs protected](/questions/logs-protected/) - [Question 37: Evidence, log retention](/questions/log-retention/) ## Sources - Microsoft cloud security benchmark PV-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-posture-vulnerability-management (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-9: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/tamper-protection/ --- # Are control-plane audit logs on in every account, subscription and project, in every region, including data access logs for sensitive stores? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 35 of the assessment asks: Are control-plane audit logs on in every account, subscription and project, in every region, including data access logs for sensitive stores? Without audit logs there is no record of who changed what, and no way to answer an auditor or an investigator. ## Why it matters Without audit logs there is no record of who changed what, and no way to answer an auditor or an investigator. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, everywhere | 1 | | In most places | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check for an organization trail covering all regions with management events, and data events for sensitive S3 buckets. - **Azure:** Check Activity Log export for all subscriptions and Entra sign-in and audit log export to a Log Analytics workspace. - **Google Cloud:** Check that Admin Activity logs are always on and that Data Access audit logs are enabled for sensitive services at the organization level. ## The fix Create an organization-wide audit trail in all regions and enable data access logs for the sensitive stores. Effort: low. ## Where it comes from - Microsoft cloud security benchmark LT-3: [Enable logging for security investigation](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection) - NIST SP 800-53 Rev. 5 AU-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 36: Evidence, logs protected](/questions/logs-protected/) - [Question 37: Evidence, log retention](/questions/log-retention/) - [Question 38: Evidence, identity alerts](/questions/identity-alerts/) ## Sources - Microsoft cloud security benchmark LT-3: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/audit-logs-everywhere/ --- # Are audit logs stored in a separate, restricted account where workload administrators cannot change or delete them? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 36 of the assessment asks: Are audit logs stored in a separate, restricted account where workload administrators cannot change or delete them? Attackers delete logs first, so logs held in the same account give weak evidence. It comes from Microsoft cloud security benchmark LT-5 and carries weight 2 in the evidence domain. ## Why it matters Attackers delete logs first, so logs held in the same account give weak evidence. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, separate and write-once | 1 | | Separate, but not write-once | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check the trail's destination bucket is in a log archive account with Object Lock and CloudTrail log file validation enabled. - **Azure:** Check that logs go to a workspace or storage in a separate subscription with immutable storage and a resource lock. - **Google Cloud:** Check log sinks to a bucket in a separate project with bucket lock or locked log buckets with a retention period. ## The fix Send logs to a dedicated security account and lock them with object lock, immutable storage or a retention lock. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark LT-5: [Centralize security log management and analysis](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection) - NIST SP 800-53 Rev. 5 AU-9: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 37: Evidence, log retention](/questions/log-retention/) - [Question 38: Evidence, identity alerts](/questions/identity-alerts/) - [Question 39: Evidence, access evidence on demand](/questions/access-evidence-on-demand/) ## Sources - Microsoft cloud security benchmark LT-5: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-9: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/logs-protected/ --- # Do you keep audit logs for at least 365 days, with the most recent 90 days searchable? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 37 of the assessment asks: Do you keep audit logs for at least 365 days, with the most recent 90 days searchable? Intrusions often run for months before they are found, so short retention removes the first weeks of the evidence. ## Why it matters Intrusions often run for months before they are found, so short retention removes the first weeks of the evidence. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, 365 days or more | 1 | | Under 365 days | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check the S3 lifecycle on the log bucket and the CloudWatch Logs retention setting for the trail's log group. - **Azure:** Check the Log Analytics workspace retention and archive settings and the storage account lifecycle policy. - **Google Cloud:** Check the retention of the _Required and _Default log buckets and of any sink destination. ## The fix Set audit log retention to at least 365 days and test a search over the oldest month. Effort: low. ## Where it comes from - Microsoft cloud security benchmark LT-6: [Configure log storage retention](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection) - NIST SP 800-53 Rev. 5 AU-11: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 38: Evidence, identity alerts](/questions/identity-alerts/) - [Question 39: Evidence, access evidence on demand](/questions/access-evidence-on-demand/) - [Question 40: Evidence, revocation drill](/questions/revocation-drill/) ## Sources - Microsoft cloud security benchmark LT-6: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-11: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/log-retention/ --- # Do alerts fire, and reach a person, for root or global administrator sign-ins, policy changes, new keys and disabled logging? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 38 of the assessment asks: Do alerts fire, and reach a person, for root or global administrator sign-ins, policy changes, new keys and disabled logging? Logs that nobody reads do not shorten an attack, while alerts on a handful of events do. ## Why it matters Logs that nobody reads do not shorten an attack, while alerts on a handful of events do. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, all four route to on-call | 1 | | Some are alerted | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Check EventBridge or CloudWatch metric filters for root login, IAM policy changes, CreateAccessKey and StopLogging, and their targets. - **Azure:** Check Azure Monitor alert rules and Entra ID Protection notifications for administrator sign-ins and role changes. - **Google Cloud:** Check Cloud Logging log-based alerts for SetIamPolicy, service account key creation and sink changes. ## The fix Create alerts for those four events and send them to the on-call channel. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark LT-2: [Enable threat detection for identity and access management](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection) - NIST SP 800-53 Rev. 5 AU-6: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 39: Evidence, access evidence on demand](/questions/access-evidence-on-demand/) - [Question 40: Evidence, revocation drill](/questions/revocation-drill/) - [Question 1: Identities and entitlements, root account mfa](/questions/root-account-mfa/) ## Sources - Microsoft cloud security benchmark LT-2: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-logging-threat-detection (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AU-6: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/identity-alerts/ --- # Within one working day, can you produce a list of who can access a given sensitive data store and how they got that access? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 39 of the assessment asks: Within one working day, can you produce a list of who can access a given sensitive data store and how they got that access? Auditors and incident responders ask this first, and the answer takes weeks when it has to be assembled by hand. ## Why it matters Auditors and incident responders ask this first, and the answer takes weeks when it has to be assembled by hand. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, within a day | 1 | | Within a few weeks | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Try IAM Access Analyzer policy generation and IAM policy simulator against one sensitive bucket, plus the identity center assignment export. - **Azure:** Try Azure role assignment export at the resource scope and Entra access package assignment reports. - **Google Cloud:** Try Policy Analyzer queries for who can access a given resource, and export the results. ## The fix Practice answering the question for one sensitive data store and write the steps down. Effort: medium. ## Where it comes from - Microsoft cloud security benchmark PA-4: [Review and reconcile user access regularly](https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access) - NIST SP 800-53 Rev. 5 AC-2: [source](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 40: Evidence, revocation drill](/questions/revocation-drill/) - [Question 1: Identities and entitlements, root account mfa](/questions/root-account-mfa/) - [Question 2: Identities and entitlements, workforce sso mfa](/questions/workforce-sso-mfa/) ## Sources - Microsoft cloud security benchmark PA-4: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-privileged-access (read 16 Nov 2026) - NIST SP 800-53 Rev. 5 AC-2: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/access-evidence-on-demand/ --- # Have you practiced cutting off all access of one compromised identity within one hour, and written down the steps? Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Question 40 of the assessment asks: Have you practiced cutting off all access of one compromised identity within one hour, and written down the steps? Revoking an identity takes many separate steps, and the first time you try it should not be during an incident. ## Why it matters Revoking an identity takes many separate steps, and the first time you try it should not be during an incident. ## Answers and what each is worth | Answer | Value | | --- | --- | | Yes, in the past 12 months | 1 | | Planned, not practiced | 0.5 | | No or not known | 0 | ## How to check it in each cloud - **AWS:** Test revoking sessions, deleting access keys and attaching a deny-all policy to one test role, then confirm CloudTrail shows no further use. - **Azure:** Test revoking sign-in sessions, disabling the user and removing role assignments for one test account, and confirm in sign-in logs. - **Google Cloud:** Test disabling the account, deleting keys and removing IAM bindings for one test principal, and confirm in audit logs. ## The fix Run a revoke-everything drill for one test identity and record how long each step takes. Effort: high. ## Where it comes from - NIST SP 800-53 Rev. 5 IR-4: [Incident Handling](https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final) ## Related pages - [Question 1: Identities and entitlements, root account mfa](/questions/root-account-mfa/) - [Question 2: Identities and entitlements, workforce sso mfa](/questions/workforce-sso-mfa/) - [Question 3: Identities and entitlements, long lived keys](/questions/long-lived-keys/) ## Sources - NIST SP 800-53 Rev. 5 IR-4: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/questions/revocation-drill/ --- # Cloud security assessment options: provider tools, open source scanners and vendor-run assessments Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The directory of assessment options: seven provider tools, six open source scanners and two vendor-run assessments, listed alphabetically within each kind. Each entry gives the maker's own description of what it checks, the clouds, the cost as published, the source and the date read. The directory ranks nothing. ## Provider tools | Name | What it checks | Clouds | Cost | Source | | --- | --- | --- | --- | --- | | [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) | AWS's documentation says it identifies resources shared with an external entity, internal access to chosen resources and unused access, validates policies against AWS best practices and generates policies from CloudTrail activity. | AWS | AWS documentation says unused access analysis is charged by the number of IAM roles and users analyzed per analyzer per month; other analyzer prices: none found on 16 Nov 2026 on the page read. | [read 16 Nov 2026](https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html) | | [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) | AWS's documentation says it runs checks against security controls from standards including AWS Foundational Security Best Practices, CIS, PCI DSS and NIST, and generates control findings. | AWS | AWS documentation says a new account gets a 30-day free trial; usage pricing is on the AWS pricing page. | [read 16 Nov 2026](https://docs.aws.amazon.com/securityhub/latest/userguide/what-is-securityhub.html) | | [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) | AWS's documentation says Trusted Advisor inspects an AWS environment and makes recommendations to save money, improve availability and performance, or close security gaps. | AWS | AWS documentation says Basic Support includes selected checks in the Security category; all checks need Business Support+, Enterprise Support or Unified Operations. | [read 16 Nov 2026](https://docs.aws.amazon.com/awssupport/latest/user/trusted-advisor.html) | | [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) | Google's documentation says role recommendations help identify and remove excess permissions and find unused service accounts. | Google Cloud | none found on 16 Nov 2026 on the page read | [read 16 Nov 2026](https://cloud.google.com/iam/docs/recommender-overview) | | [Google Security Command Center](/options/google-security-command-center/) | Google's documentation says it is a cloud-based risk management solution that helps security professionals prevent, detect and respond to security issues, including misconfigurations and publicly exposed resources. | Google Cloud, and AWS through connectors that Google documents | Google's pricing page lists three tiers, Standard, Premium and Enterprise, and says Standard is free of charge. | [read 16 Nov 2026](https://cloud.google.com/security-command-center/docs/concepts-security-command-center-overview) | | [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) | Microsoft's documentation says the secure score aggregates security findings into a single score and that a higher score means a lower identified risk level, based on recommendations from the Microsoft cloud security benchmark. | Azure, and AWS and Google Cloud through connectors (Microsoft documentation describes multicloud environments) | Microsoft's documentation says Foundational CSPM is available at no cost and Defender CSPM is a paid plan. | [read 16 Nov 2026](https://learn.microsoft.com/en-us/azure/defender-for-cloud/secure-score-security-controls) | | [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) | Microsoft's documentation says the Microsoft Entra recommendations feature helps monitor the status of a tenant so that it stays in a secure and healthy state. | Microsoft Entra ID (Azure) | Microsoft's documentation says license requirements depend on the recommendation and that some need a P2 or other license. | [read 16 Nov 2026](https://learn.microsoft.com/en-us/entra/identity/monitoring-health/overview-recommendations) | ## Open source scanners | Name | What it checks | Clouds | Cost | Source | | --- | --- | --- | --- | --- | | [Cartography](/options/cartography/) | Its GitHub page says Cartography is a Python tool that pulls infrastructure assets and their relationships into a Neo4j graph database. | AWS, Google Cloud, Azure and more than 30 other platforms, as listed on its GitHub page | Apache-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/cartography-cncf/cartography) | | [CloudSploit](/options/cloudsploit/) | Its GitHub page describes it as Cloud Security Posture Management (CSPM) and lists AWS, Azure, Google Cloud, Oracle and Alibaba Cloud among its topics. | AWS, Azure, Google Cloud, Oracle Cloud, Alibaba Cloud (GitHub topics) | GPL-3.0 license shown on its GitHub page; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/aquasecurity/cloudsploit) | | [PMapper](/options/pmapper/) | Its GitHub page says Principal Mapper identifies risks in the configuration of AWS IAM for an account or organization, modeling users and roles as a directed graph to check for privilege escalation. | AWS | AGPL-3.0 license shown on its GitHub page; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/nccgroup/PMapper) | | [Prowler](/options/prowler/) | Its GitHub page says Prowler automates security and compliance checks and compliance frameworks across cloud environments. | AWS, Azure, Google Cloud and others (GitHub topics) | Apache-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/prowler-cloud/prowler) | | [ScoutSuite](/options/scoutsuite/) | Its GitHub page says Scout Suite is an open source multi-cloud security-auditing tool that gathers configuration data through provider APIs and highlights risk areas. | AWS, Azure, Google Cloud (GitHub topics) | GPL-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/nccgroup/ScoutSuite) | | [Steampipe compliance mods](/options/steampipe-compliance-mods/) | Its GitHub page says the AWS compliance mod runs individual controls or full compliance benchmarks for CIS, PCI, NIST, HIPAA and more across AWS accounts using Powerpipe and Steampipe. | AWS (this mod); Steampipe plugins also cover Azure and Google Cloud | License shown on its GitHub page without a stated type; price: none found on 16 Nov 2026 | [read 16 Nov 2026](https://github.com/turbot/steampipe-mod-aws-compliance) | ## Vendor-run assessments | Name | What it checks | Clouds | Cost | Source | | --- | --- | --- | --- | --- | | [Act free assessment](/options/act-free-assessment/) | Act's page says it is a free access attack simulation that is read-only, deployed in under 5 minutes, and confirms 5 real attack scenarios in your environment covering external trust, data exfiltration, AI agents, vibecoding and cross-environment. | The page says cloud environment; the individual clouds: none found on 16 Nov 2026 | Act describes it as free; other terms: none found on 16 Nov 2026 | [read 16 Nov 2026](https://act.security/free-assessment?utm_source=cloudsecurityriskassessment.com&utm_medium=referral&utm_campaign=directory) | | [Wiz Cloud Security Self Assessment](/options/wiz-cloud-security-self-assessment/) | Wiz's page says users answer 9 introductory questions across 9 security domains, get detailed checklists for domains answered yes, and receive a risk score with a breakdown by domain. | Clouds: none found on 16 Nov 2026 on the page read | none found on 16 Nov 2026 | [read 16 Nov 2026](https://www.wiz.io/cloud-security-assessment) | ## Related pages - [Guides](/guides/) - [Glossary](/glossary/) - [News](/news/) ## Sources - AWS IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) - AWS Security Hub CSPM: https://docs.aws.amazon.com/securityhub/latest/userguide/what-is-securityhub.html (read 16 Nov 2026) - AWS Trusted Advisor security checks: https://docs.aws.amazon.com/awssupport/latest/user/trusted-advisor.html (read 16 Nov 2026) - Google Cloud IAM recommender: https://cloud.google.com/iam/docs/recommender-overview (read 16 Nov 2026) - Google Security Command Center: https://cloud.google.com/security-command-center/docs/concepts-security-command-center-overview (read 16 Nov 2026) - Microsoft Defender for Cloud secure score: https://learn.microsoft.com/en-us/azure/defender-for-cloud/secure-score-security-controls (read 16 Nov 2026) - Microsoft Entra ID recommendations: https://learn.microsoft.com/en-us/entra/identity/monitoring-health/overview-recommendations (read 16 Nov 2026) - Cartography: https://github.com/cartography-cncf/cartography (read 16 Nov 2026) - CloudSploit: https://github.com/aquasecurity/cloudsploit (read 16 Nov 2026) - PMapper: https://github.com/nccgroup/PMapper (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/ --- # AWS IAM Access Analyzer as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS IAM Access Analyzer is listed among the provider tools in the directory of cloud security assessment options. AWS's documentation says it identifies resources shared with an external entity, internal access to chosen resources and unused access, validates policies against AWS best practices and generates policies from CloudTrail activity. - **Name:** AWS IAM Access Analyzer - **Kind:** Provider tool - **What it checks:** AWS's documentation says it identifies resources shared with an external entity, internal access to chosen resources and unused access, validates policies against AWS best practices and generates policies from CloudTrail activity. - **Clouds:** AWS - **Cost:** AWS documentation says unused access analysis is charged by the number of IAM roles and users analyzed per analyzer per month; other analyzer prices: none found on 16 Nov 2026 on the page read. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; AWS IAM Access Analyzer covers part of that ground in its own way, as its maker describes it above, and AWS IAM Access Analyzer can run beside the assessment. ## Related pages - [AWS Security Hub CSPM: provider tool for cloud assessment](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks: provider tool](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender: provider tool](/options/google-cloud-iam-recommender/) ## Sources - AWS IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/aws-iam-access-analyzer/ --- # AWS Security Hub CSPM as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS Security Hub CSPM is listed among the provider tools in the directory of cloud security assessment options. AWS's documentation says it runs checks against security controls from standards including AWS Foundational Security Best Practices, CIS, PCI DSS and NIST, and generates control findings. - **Name:** AWS Security Hub CSPM - **Kind:** Provider tool - **What it checks:** AWS's documentation says it runs checks against security controls from standards including AWS Foundational Security Best Practices, CIS, PCI DSS and NIST, and generates control findings. - **Clouds:** AWS - **Cost:** AWS documentation says a new account gets a 30-day free trial; usage pricing is on the AWS pricing page. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; AWS Security Hub CSPM covers part of that ground in its own way, as its maker describes it above, and AWS Security Hub CSPM can run beside the assessment. ## Related pages - [AWS Trusted Advisor security checks: provider tool](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender: provider tool](/options/google-cloud-iam-recommender/) - [Google Security Command Center: provider tool](/options/google-security-command-center/) ## Sources - AWS Security Hub CSPM: https://docs.aws.amazon.com/securityhub/latest/userguide/what-is-securityhub.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/aws-security-hub-cspm/ --- # AWS Trusted Advisor security checks as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS Trusted Advisor security checks is listed among the provider tools in the directory of cloud security assessment options. AWS's documentation says Trusted Advisor inspects an AWS environment and makes recommendations to save money, improve availability and performance, or close security gaps. - **Name:** AWS Trusted Advisor security checks - **Kind:** Provider tool - **What it checks:** AWS's documentation says Trusted Advisor inspects an AWS environment and makes recommendations to save money, improve availability and performance, or close security gaps. - **Clouds:** AWS - **Cost:** AWS documentation says Basic Support includes selected checks in the Security category; all checks need Business Support+, Enterprise Support or Unified Operations. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; AWS Trusted Advisor security checks covers part of that ground in its own way, as its maker describes it above, and AWS Trusted Advisor security checks can run beside the assessment. ## Related pages - [Google Cloud IAM recommender: provider tool](/options/google-cloud-iam-recommender/) - [Google Security Command Center: provider tool](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score: provider tool](/options/microsoft-defender-secure-score/) ## Sources - AWS Trusted Advisor security checks: https://docs.aws.amazon.com/awssupport/latest/user/trusted-advisor.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/aws-trusted-advisor/ --- # Google Cloud IAM recommender as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Google Cloud IAM recommender is listed among the provider tools in the directory of cloud security assessment options. Google's documentation says role recommendations help identify and remove excess permissions and find unused service accounts. Clouds: Google Cloud. Cost: none found on 16 Nov 2026 on the page read. - **Name:** Google Cloud IAM recommender - **Kind:** Provider tool - **What it checks:** Google's documentation says role recommendations help identify and remove excess permissions and find unused service accounts. - **Clouds:** Google Cloud - **Cost:** none found on 16 Nov 2026 on the page read - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Google Cloud IAM recommender covers part of that ground in its own way, as its maker describes it above, and Google Cloud IAM recommender can run beside the assessment. ## Related pages - [Google Security Command Center: provider tool](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score: provider tool](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations: provider tool](/options/microsoft-entra-recommendations/) ## Sources - Google Cloud IAM recommender: https://cloud.google.com/iam/docs/recommender-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/google-cloud-iam-recommender/ --- # Google Security Command Center as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Google Security Command Center is listed among the provider tools in the directory of cloud security assessment options. Google's documentation says it is a cloud-based risk management solution that helps security professionals prevent, detect and respond to security issues, including misconfigurations and publicly exposed resources. - **Name:** Google Security Command Center - **Kind:** Provider tool - **What it checks:** Google's documentation says it is a cloud-based risk management solution that helps security professionals prevent, detect and respond to security issues, including misconfigurations and publicly exposed resources. - **Clouds:** Google Cloud, and AWS through connectors that Google documents - **Cost:** Google's pricing page lists three tiers, Standard, Premium and Enterprise, and says Standard is free of charge. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Google Security Command Center covers part of that ground in its own way, as its maker describes it above, and Google Security Command Center can run beside the assessment. ## Related pages - [Microsoft Defender for Cloud secure score: provider tool](/options/microsoft-defender-secure-score/) - [Microsoft Entra ID recommendations: provider tool](/options/microsoft-entra-recommendations/) - [Cartography: open source scanner for cloud assessment](/options/cartography/) ## Sources - Google Security Command Center: https://cloud.google.com/security-command-center/docs/concepts-security-command-center-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/google-security-command-center/ --- # Microsoft Defender for Cloud secure score as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Microsoft Defender for Cloud secure score is listed among the provider tools in the directory of cloud security assessment options. Microsoft's documentation says the secure score aggregates security findings into a single score and that a higher score means a lower identified risk level, based on recommendations from the Microsoft cloud security benchmark. - **Name:** Microsoft Defender for Cloud secure score - **Kind:** Provider tool - **What it checks:** Microsoft's documentation says the secure score aggregates security findings into a single score and that a higher score means a lower identified risk level, based on recommendations from the Microsoft cloud security benchmark. - **Clouds:** Azure, and AWS and Google Cloud through connectors (Microsoft documentation describes multicloud environments) - **Cost:** Microsoft's documentation says Foundational CSPM is available at no cost and Defender CSPM is a paid plan. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Entra ID recommendations](/options/microsoft-entra-recommendations/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Microsoft Defender for Cloud secure score covers part of that ground in its own way, as its maker describes it above, and Microsoft Defender for Cloud secure score can run beside the assessment. ## Related pages - [Microsoft Entra ID recommendations: provider tool](/options/microsoft-entra-recommendations/) - [Cartography: open source scanner for cloud assessment](/options/cartography/) - [CloudSploit: open source scanner for cloud assessment](/options/cloudsploit/) ## Sources - Microsoft Defender for Cloud secure score: https://learn.microsoft.com/en-us/azure/defender-for-cloud/secure-score-security-controls (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/microsoft-defender-secure-score/ --- # Microsoft Entra ID recommendations as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Microsoft Entra ID recommendations is listed among the provider tools in the directory of cloud security assessment options. Microsoft's documentation says the Microsoft Entra recommendations feature helps monitor the status of a tenant so that it stays in a secure and healthy state. - **Name:** Microsoft Entra ID recommendations - **Kind:** Provider tool - **What it checks:** Microsoft's documentation says the Microsoft Entra recommendations feature helps monitor the status of a tenant so that it stays in a secure and healthy state. - **Clouds:** Microsoft Entra ID (Azure) - **Cost:** Microsoft's documentation says license requirements depend on the recommendation and that some need a P2 or other license. - **Read:** 16 Nov 2026 ## Other provider tools - [AWS IAM Access Analyzer](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks](/options/aws-trusted-advisor/) - [Google Cloud IAM recommender](/options/google-cloud-iam-recommender/) - [Google Security Command Center](/options/google-security-command-center/) - [Microsoft Defender for Cloud secure score](/options/microsoft-defender-secure-score/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Microsoft Entra ID recommendations covers part of that ground in its own way, as its maker describes it above, and Microsoft Entra ID recommendations can run beside the assessment. ## Related pages - [Cartography: open source scanner for cloud assessment](/options/cartography/) - [CloudSploit: open source scanner for cloud assessment](/options/cloudsploit/) - [PMapper: open source scanner for cloud assessment](/options/pmapper/) ## Sources - Microsoft Entra ID recommendations: https://learn.microsoft.com/en-us/entra/identity/monitoring-health/overview-recommendations (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/microsoft-entra-recommendations/ --- # Cartography as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cartography is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page says Cartography is a Python tool that pulls infrastructure assets and their relationships into a Neo4j graph database. Clouds: AWS, Google Cloud, Azure and more than 30 other platforms, as listed on its GitHub page. - **Name:** Cartography - **Kind:** Open source scanner - **What it checks:** Its GitHub page says Cartography is a Python tool that pulls infrastructure assets and their relationships into a Neo4j graph database. - **Clouds:** AWS, Google Cloud, Azure and more than 30 other platforms, as listed on its GitHub page - **Cost:** Apache-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [CloudSploit](/options/cloudsploit/) - [PMapper](/options/pmapper/) - [Prowler](/options/prowler/) - [ScoutSuite](/options/scoutsuite/) - [Steampipe compliance mods](/options/steampipe-compliance-mods/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Cartography covers part of that ground in its own way, as its maker describes it above, and Cartography can run beside the assessment. ## Related pages - [CloudSploit: open source scanner for cloud assessment](/options/cloudsploit/) - [PMapper: open source scanner for cloud assessment](/options/pmapper/) - [Prowler: open source scanner for cloud assessment](/options/prowler/) ## Sources - Cartography: https://github.com/cartography-cncf/cartography (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/cartography/ --- # CloudSploit as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. CloudSploit is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page describes it as Cloud Security Posture Management (CSPM) and lists AWS, Azure, Google Cloud, Oracle and Alibaba Cloud among its topics. - **Name:** CloudSploit - **Kind:** Open source scanner - **What it checks:** Its GitHub page describes it as Cloud Security Posture Management (CSPM) and lists AWS, Azure, Google Cloud, Oracle and Alibaba Cloud among its topics. - **Clouds:** AWS, Azure, Google Cloud, Oracle Cloud, Alibaba Cloud (GitHub topics) - **Cost:** GPL-3.0 license shown on its GitHub page; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [Cartography](/options/cartography/) - [PMapper](/options/pmapper/) - [Prowler](/options/prowler/) - [ScoutSuite](/options/scoutsuite/) - [Steampipe compliance mods](/options/steampipe-compliance-mods/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; CloudSploit covers part of that ground in its own way, as its maker describes it above, and CloudSploit can run beside the assessment. ## Related pages - [PMapper: open source scanner for cloud assessment](/options/pmapper/) - [Prowler: open source scanner for cloud assessment](/options/prowler/) - [ScoutSuite: open source scanner for cloud assessment](/options/scoutsuite/) ## Sources - CloudSploit: https://github.com/aquasecurity/cloudsploit (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/cloudsploit/ --- # PMapper as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. PMapper is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page says Principal Mapper identifies risks in the configuration of AWS IAM for an account or organization, modeling users and roles as a directed graph to check for privilege escalation. - **Name:** PMapper - **Kind:** Open source scanner - **What it checks:** Its GitHub page says Principal Mapper identifies risks in the configuration of AWS IAM for an account or organization, modeling users and roles as a directed graph to check for privilege escalation. - **Clouds:** AWS - **Cost:** AGPL-3.0 license shown on its GitHub page; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [Cartography](/options/cartography/) - [CloudSploit](/options/cloudsploit/) - [Prowler](/options/prowler/) - [ScoutSuite](/options/scoutsuite/) - [Steampipe compliance mods](/options/steampipe-compliance-mods/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; PMapper covers part of that ground in its own way, as its maker describes it above, and PMapper can run beside the assessment. ## Related pages - [Prowler: open source scanner for cloud assessment](/options/prowler/) - [ScoutSuite: open source scanner for cloud assessment](/options/scoutsuite/) - [Steampipe compliance mods: open source scanner](/options/steampipe-compliance-mods/) ## Sources - PMapper: https://github.com/nccgroup/PMapper (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/pmapper/ --- # Prowler as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Prowler is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page says Prowler automates security and compliance checks and compliance frameworks across cloud environments. Clouds: AWS, Azure, Google Cloud and others (GitHub topics). - **Name:** Prowler - **Kind:** Open source scanner - **What it checks:** Its GitHub page says Prowler automates security and compliance checks and compliance frameworks across cloud environments. - **Clouds:** AWS, Azure, Google Cloud and others (GitHub topics) - **Cost:** Apache-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [Cartography](/options/cartography/) - [CloudSploit](/options/cloudsploit/) - [PMapper](/options/pmapper/) - [ScoutSuite](/options/scoutsuite/) - [Steampipe compliance mods](/options/steampipe-compliance-mods/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Prowler covers part of that ground in its own way, as its maker describes it above, and Prowler can run beside the assessment. ## Related pages - [ScoutSuite: open source scanner for cloud assessment](/options/scoutsuite/) - [Steampipe compliance mods: open source scanner](/options/steampipe-compliance-mods/) - [Act free assessment: vendor-run assessment](/options/act-free-assessment/) ## Sources - Prowler: https://github.com/prowler-cloud/prowler (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/prowler/ --- # ScoutSuite as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. ScoutSuite is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page says Scout Suite is an open source multi-cloud security-auditing tool that gathers configuration data through provider APIs and highlights risk areas. - **Name:** ScoutSuite - **Kind:** Open source scanner - **What it checks:** Its GitHub page says Scout Suite is an open source multi-cloud security-auditing tool that gathers configuration data through provider APIs and highlights risk areas. - **Clouds:** AWS, Azure, Google Cloud (GitHub topics) - **Cost:** GPL-2.0 license shown on its GitHub page; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [Cartography](/options/cartography/) - [CloudSploit](/options/cloudsploit/) - [PMapper](/options/pmapper/) - [Prowler](/options/prowler/) - [Steampipe compliance mods](/options/steampipe-compliance-mods/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; ScoutSuite covers part of that ground in its own way, as its maker describes it above, and ScoutSuite can run beside the assessment. ## Related pages - [Steampipe compliance mods: open source scanner](/options/steampipe-compliance-mods/) - [Act free assessment: vendor-run assessment](/options/act-free-assessment/) - [Wiz Cloud Security Self Assessment: vendor-run assessment](/options/wiz-cloud-security-self-assessment/) ## Sources - ScoutSuite: https://github.com/nccgroup/ScoutSuite (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/scoutsuite/ --- # Steampipe compliance mods as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Steampipe compliance mods is listed among the open source scanners in the directory of cloud security assessment options. Its GitHub page says the AWS compliance mod runs individual controls or full compliance benchmarks for CIS, PCI, NIST, HIPAA and more across AWS accounts using Powerpipe and Steampipe. - **Name:** Steampipe compliance mods - **Kind:** Open source scanner - **What it checks:** Its GitHub page says the AWS compliance mod runs individual controls or full compliance benchmarks for CIS, PCI, NIST, HIPAA and more across AWS accounts using Powerpipe and Steampipe. - **Clouds:** AWS (this mod); Steampipe plugins also cover Azure and Google Cloud - **Cost:** License shown on its GitHub page without a stated type; price: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other open source scanners - [Cartography](/options/cartography/) - [CloudSploit](/options/cloudsploit/) - [PMapper](/options/pmapper/) - [Prowler](/options/prowler/) - [ScoutSuite](/options/scoutsuite/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Steampipe compliance mods covers part of that ground in its own way, as its maker describes it above, and Steampipe compliance mods can run beside the assessment. ## Related pages - [Act free assessment: vendor-run assessment](/options/act-free-assessment/) - [Wiz Cloud Security Self Assessment: vendor-run assessment](/options/wiz-cloud-security-self-assessment/) - [AWS IAM Access Analyzer: provider tool](/options/aws-iam-access-analyzer/) ## Sources - Steampipe compliance mods: https://github.com/turbot/steampipe-mod-aws-compliance (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/steampipe-compliance-mods/ --- # Act free assessment as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Act free assessment is listed among the vendor-run assessments in the directory of cloud security assessment options. Act's page says it is a free access attack simulation that is read-only, deployed in under 5 minutes, and confirms 5 real attack scenarios in your environment covering external trust, data exfiltration, AI agents, vibecoding and cross-environment. - **Name:** Act free assessment - **Kind:** Vendor-run assessment - **What it checks:** Act's page says it is a free access attack simulation that is read-only, deployed in under 5 minutes, and confirms 5 real attack scenarios in your environment covering external trust, data exfiltration, AI agents, vibecoding and cross-environment. - **Clouds:** The page says cloud environment; the individual clouds: none found on 16 Nov 2026 - **Cost:** Act describes it as free; other terms: none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other vendor-run assessments - [Wiz Cloud Security Self Assessment](/options/wiz-cloud-security-self-assessment/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Act free assessment covers part of that ground in its own way, as its maker describes it above, and Act free assessment can run beside the assessment. ## Related pages - [Wiz Cloud Security Self Assessment: vendor-run assessment](/options/wiz-cloud-security-self-assessment/) - [AWS IAM Access Analyzer: provider tool](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM: provider tool for cloud assessment](/options/aws-security-hub-cspm/) ## Sources - Act free assessment: https://act.security/free-assessment (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/act-free-assessment/ --- # Wiz Cloud Security Self Assessment as a cloud security assessment option Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Wiz Cloud Security Self Assessment is listed among the vendor-run assessments in the directory of cloud security assessment options. Wiz's page says users answer 9 introductory questions across 9 security domains, get detailed checklists for domains answered yes, and receive a risk score with a breakdown by domain. - **Name:** Wiz Cloud Security Self Assessment - **Kind:** Vendor-run assessment - **What it checks:** Wiz's page says users answer 9 introductory questions across 9 security domains, get detailed checklists for domains answered yes, and receive a risk score with a breakdown by domain. - **Clouds:** Clouds: none found on 16 Nov 2026 on the page read - **Cost:** none found on 16 Nov 2026 - **Read:** 16 Nov 2026 ## Other vendor-run assessments - [Act free assessment](/options/act-free-assessment/) ## Where it fits The [assessment](/) scores who and what can reach your cloud resources across six domains; Wiz Cloud Security Self Assessment covers part of that ground in its own way, as its maker describes it above, and Wiz Cloud Security Self Assessment can run beside the assessment. ## Related pages - [AWS IAM Access Analyzer: provider tool](/options/aws-iam-access-analyzer/) - [AWS Security Hub CSPM: provider tool for cloud assessment](/options/aws-security-hub-cspm/) - [AWS Trusted Advisor security checks: provider tool](/options/aws-trusted-advisor/) ## Sources - Wiz Cloud Security Self Assessment: https://www.wiz.io/cloud-security-assessment (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/options/wiz-cloud-security-self-assessment/ --- # A cloud access risk register, one row per question Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The risk register is an XLSX made in your browser from the assessment answers. It has 15 columns, from check ID and weight to gap, owner and due date, and gives every gap an owner and a recheck date. The register template guide explains each column and how to keep the sheet current. | Column | What goes in it | | --- | --- | | Id, Risk, Domain, Control source | Filled in from the 40 questions | | Likelihood, Impact | 1 to 5 each, your judgment | | Score | Likelihood times impact | | Owner, Due date, Status | Yours to fill | | Fix, Effort | From the question page | ## Related pages - [30 questions about cloud access risk assessment](/faq/) - [The 40-question assessment: the data, downloadable](/datasets/) - [About](/about/) ## Sources - CIS AWS Foundations Benchmark v5.0: https://www.cisecurity.org/benchmark/amazon_web_services (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/risk-register/ --- # Guides Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Guides to assessing the cloud access layer: checklists for the assessment, AWS, Azure, Google Cloud and audit use, the maturity model, the risk assessment method, the audit guide, a register template and a 15-item list for a new account. Each guide states its scope, links to the questions and lists sources with read dates. - [Cloud Security Assessment Checklist: 40 Checks](/cloud-security-assessment-checklist/): All 40 checks grouped by domain, with a Word download. - [AWS Security Checklist for the Access Layer](/aws-security-checklist/): Twelve checks with the AWS place to look. - [Azure Security Checklist for the Access Layer](/azure-security-checklist/): Twelve checks with the Azure and Entra place to look. - [Google Cloud Security Checklist: Access Layer](/google-cloud-security-checklist/): Twelve checks with the Google Cloud place to look. - [Cloud Security Audit Checklist: Evidence to Collect](/cloud-security-audit-checklist/): Evidence to collect for each of the six domains, with a Word download. - [Cloud Security Audit: What It Covers and Proves](/cloud-security-audit/): What an access-layer audit covers, who runs it and what it produces. - [Cloud Security Maturity Model: Five Levels](/cloud-security-maturity-model/): Five levels, each with a score band and observable criteria. - [Cloud Risk Assessment: Score Gaps by Weight](/cloud-risk-assessment/): The scoring arithmetic with a worked example. - [Cloud Security Assessment: Scope and Time](/cloud-security-assessment/): Scope, time and outputs of an access-layer assessment. - [AWS Security Assessment: Steps for the Access Layer](/aws-security-assessment/): Seven steps with the AWS service for each. - [Cloud Risk Register Template (XLSX): Columns](/risk-register-template/): Columns, the gap score and who owns each row. - [Cloud Security Checklist: 15 Items for a New Account](/cloud-security-checklist/): A 15-item list for a new account, distinct from the 40-check list. ## Related pages - [Glossary](/glossary/) - [News](/news/) - [The 40 questions of the assessment](/questions/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/guides/ --- # Cloud security assessment checklist: 40 checks for the cloud access layer Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The cloud security assessment checklist lists 40 checks in six domains: identities and entitlements (8), network reachability (7), data perimeter (7), AI workloads (6), guardrails and drift (6) and evidence (6). Each applies to AWS, Azure and Google Cloud and weighs 1 to 3 in the score. Edition 16 Nov 2026. ## Forty checks sit in six domains, and the domains carry unequal weight The checklist has 40 checks. Each is a yes, partly or no question about the cloud access layer, and each applies to AWS, Azure and Google Cloud. A check weighs 1, 2 or 3 depending on how far a wrong answer moves risk. The 40 weights add up to 84, and 12 checks weigh 3. Domain weights are set apart from the check weights: identities and entitlements 25, network reachability 20, data perimeter 20, AI workloads 15, guardrails and drift 10, evidence 10. The domain weights add up to 100. | Domain | Checks | Domain weight | Sum of check weights | | --- | --- | --- | --- | | Identities and entitlements | 8 | 25 | 19 | | Network reachability | 7 | 20 | 14 | | Data perimeter | 7 | 20 | 15 | | AI workloads | 6 | 15 | 14 | | Guardrails and drift | 6 | 10 | 11 | | Evidence | 6 | 10 | 11 | ## Eight identity checks cover who holds power and for how long People and workloads that hold standing power over the cloud account are the most common route from one stolen credential to a full breach. - [Root or global administrator has phishing-resistant MFA and no keys](/questions/root-account-mfa/) (weight 3, fix effort low) - [People sign in through single sign-on with MFA](/questions/workforce-sso-mfa/) (weight 3, fix effort medium) - [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/) (weight 3, fix effort medium) - [Workloads use platform-issued identities](/questions/workload-identity/) (weight 2, fix effort high) - [Administrator roles are few, named and separate](/questions/admin-roles-limited/) (weight 2, fix effort low) - [Privileged roles are granted for a limited time](/questions/no-standing-privilege/) (weight 3, fix effort high) - [Permissions reviewed every 90 days](/questions/access-reviews/) (weight 2, fix effort medium) - [Emergency accounts tested twice a year](/questions/emergency-access/) (weight 1, fix effort low) ## Seven network checks decide what an attacker can touch without a credential Reachability decides which of your services an attacker can touch before any credential is needed. - [No admin or database port open to the internet](/questions/no-open-admin-ports/) (weight 3, fix effort low) - [Data services reached through private endpoints](/questions/private-service-endpoints/) (weight 2, fix effort high) - [No public IP address by default](/questions/no-default-public-ip/) (weight 2, fix effort medium) - [Admin access brokered and logged](/questions/brokered-admin-access/) (weight 2, fix effort medium) - [Metadata service hardened](/questions/metadata-service-hardened/) (weight 3, fix effort low) - [Outbound traffic filtered to known destinations](/questions/egress-controlled/) (weight 1, fix effort high) - [Reachability analyzed monthly](/questions/reachability-analysis/) (weight 1, fix effort medium) ## Seven data perimeter checks limit who reads data and from where A data perimeter limits who can reach your data and from where, even when an identity or a policy is wrong. - [Public object storage blocked organization-wide](/questions/public-storage-blocked/) (weight 3, fix effort low) - [Sharing limited to your own organization](/questions/organization-boundary/) (weight 3, fix effort medium) - [Sensitive data readable from trusted networks only](/questions/trusted-network-condition/) (weight 2, fix effort high) - [External access findings reviewed weekly](/questions/external-access-findings/) (weight 2, fix effort low) - [Customer-managed keys with limited use](/questions/key-control/) (weight 2, fix effort medium) - [Sensitive data locations known](/questions/sensitive-data-located/) (weight 1, fix effort medium) - [Third-party roles conditioned, owned and dated](/questions/third-party-access/) (weight 2, fix effort medium) ## Six AI checks treat each agent as an identity with a reach AI services and agents act with cloud credentials, so what they can call and read is an access question. - [Each AI agent has its own identity](/questions/ai-own-identity/) (weight 3, fix effort medium) - [Agent actions limited to the task and written down](/questions/ai-tool-permissions/) (weight 3, fix effort high) - [Only approved models can be used](/questions/approved-models-only/) (weight 2, fix effort low) - [AI data reach matches user rights](/questions/ai-data-reach/) (weight 2, fix effort high) - [Human approval for high-impact agent actions](/questions/human-approval/) (weight 2, fix effort medium) - [Model and tool calls logged with the identity](/questions/ai-activity-logging/) (weight 2, fix effort low) ## Six guardrail checks stop bad changes and catch the ones that slip past Preventive controls stop a bad change before it lands, and drift checks catch the ones that slip past. - [Deny policies at the organization level](/questions/preventive-policies/) (weight 3, fix effort medium) - [Changes go through code and pipelines](/questions/changes-as-code/) (weight 2, fix effort high) - [Drift reported within a day](/questions/drift-detection/) (weight 2, fix effort medium) - [Guardrails tested before enforcement](/questions/guardrail-testing/) (weight 1, fix effort medium) - [Exceptions have owner, reason and expiry](/questions/exceptions-expire/) (weight 1, fix effort low) - [Guardrails protected from tampering](/questions/tamper-protection/) (weight 2, fix effort medium) ## Six evidence checks let you prove access and cut it off Logs and records are what let you prove who had access, detect misuse and respond. - [Audit logs on in every account and region](/questions/audit-logs-everywhere/) (weight 3, fix effort low) - [Logs stored where workload admins cannot change them](/questions/logs-protected/) (weight 2, fix effort medium) - [Logs kept 365 days, 90 searchable](/questions/log-retention/) (weight 2, fix effort low) - [Identity alerts reach a person](/questions/identity-alerts/) (weight 2, fix effort medium) - [Access evidence produced in one working day](/questions/access-evidence-on-demand/) (weight 1, fix effort medium) - [One identity cut off within one hour](/questions/revocation-drill/) (weight 1, fix effort high) ## The checklist is also a Word file you can edit The Word version, [cloud-security-assessment-checklist](/cloud-security-assessment-checklist/), lists the same 40 checks with columns for the answer, the owner and the evidence. Use it when a team wants to review the checks in a meeting before anyone opens the scored tool. To get a score, answer the same checks in the [assessment](/). The scoring is explained in the [methodology](/methodology/), and the [risk register template](/risk-register-template/) turns the five fixes into tracked work. Providers publish more checks than these 40, as the [options](/options/) page shows. ## Each check names the control it comes from Every check cites one primary control: a CIS Benchmark, the Microsoft cloud security benchmark, a provider guide or NIST SP 800-53 Rev. 5. CIS says it publishes more than 100 Benchmarks across 25 or more vendor product families (CIS Benchmarks page, read 16 Nov 2026). The checklist refers to CIS and CSA controls by number and short title only, because those texts are licensed. ## Download - [Cloud security assessment checklist (.docx)](/downloads/cloud-security-assessment-checklist.docx): Word document made from this page, edition 16 Nov 2026 ## Related pages - [AWS Security Checklist for the Access Layer](/aws-security-checklist/) - [Azure Security Checklist for the Access Layer](/azure-security-checklist/) - [Google Cloud Security Checklist: Access Layer](/google-cloud-security-checklist/) - [All of them](/guides/) ## Sources - CIS Benchmarks: https://www.cisecurity.org/cis-benchmarks (read 16 Nov 2026) - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - Microsoft cloud security benchmark overview: https://learn.microsoft.com/en-us/security/benchmark/azure/overview (read 16 Nov 2026) - AWS Well-Architected Security Pillar: https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/welcome.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-assessment-checklist/ --- # AWS security checklist: 12 access checks to run in every account Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The AWS security checklist here picks 12 of the 40 access-layer checks and says where to look in AWS for each: the IAM credential report, security groups, S3 Block Public Access, service control policies, IAM Access Analyzer and CloudTrail. Run them in every account of the organization. Read 16 Nov 2026. ## Twelve checks cover the AWS access layer The checks below come from the 40-check [assessment checklist](/cloud-security-assessment-checklist/). They are the ones with a direct AWS console or CLI test. Run each in every account of the organization, not only the production account, because one forgotten account is enough for a path in. | Check | Where to look in AWS | | --- | --- | | [Root or global administrator has phishing-resistant MFA and no keys](/questions/root-account-mfa/) | In the IAM console credential report, check that the root row shows mfa_active true and access_key_1_active and access_key_2_active false, in every account of the organization. | | [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/) | Download the IAM credential report and sort by access_key_last_rotated and access_key_last_used; every active key older than 90 days is a finding. | | [Workloads use platform-issued identities](/questions/workload-identity/) | Check that EC2, ECS, EKS and Lambda use execution or instance roles and that CI systems use OIDC federation to assume roles instead of stored access keys. | | [No admin or database port open to the internet](/questions/no-open-admin-ports/) | Describe security groups with ip-permission.cidr 0.0.0.0/0 on ports 22, 3389, 3306, 5432 and 1433, or use the matching AWS Config managed rules. | | [Metadata service hardened](/questions/metadata-service-hardened/) | Describe instances and check MetadataOptions.HttpTokens is required, and that the account default under EC2 data protection and security is set to required. | | [Public object storage blocked organization-wide](/questions/public-storage-blocked/) | Check S3 Block Public Access at the account level in every account, or the organization-level policy, with all four settings on. | | [Sharing limited to your own organization](/questions/organization-boundary/) | Check SCPs or resource control policies for aws:PrincipalOrgID and aws:ResourceOrgID conditions, and test with Access Analyzer external access findings. | | [External access findings reviewed weekly](/questions/external-access-findings/) | Check that an IAM Access Analyzer for the organization exists with external access type, and count active findings and their age. | | [Deny policies at the organization level](/questions/preventive-policies/) | List SCPs attached to the organization root and OUs (aws organizations list-policies) and check they deny actions such as disabling logging or leaving the organization. | | [Guardrails protected from tampering](/questions/tamper-protection/) | Check SCPs deny cloudtrail:StopLogging, config:StopConfigurationRecorder and guardduty:DeleteDetector except for a security role. | | [Audit logs on in every account and region](/questions/audit-logs-everywhere/) | Check for an organization trail covering all regions with management events, and data events for sensitive S3 buckets. | | [Logs stored where workload admins cannot change them](/questions/logs-protected/) | Check the trail's destination bucket is in a log archive account with Object Lock and CloudTrail log file validation enabled. | ## The CIS AWS Foundations Benchmark is the reference for the root and logging checks The first check cites the CIS AWS Foundations Benchmark (controls 1.4 and 1.5 in the edition the tool cites). CIS announced AWS Foundations Benchmark v7.1.0 in its November 2026 update, and CIS lists expanded AWS Organizations and IAM coverage, enhanced access and credential management guidance and a new section on AWS Bedrock security and governance as highlights (CIS, read 16 Nov 2026). Check the control numbers against the edition you hold before you cite them in an audit. The tool refers to CIS controls by number and short title only. ## Three AWS services give most of the evidence The IAM credential report shows root MFA, key age and key use for every user in an account, which answers [root account MFA](/questions/root-account-mfa/) and [long-lived keys](/questions/long-lived-keys/) in one file. AWS describes IAM Access Analyzer as finding resources shared with an external entity, internal access to chosen resources and unused access (AWS documentation, read 16 Nov 2026). AWS Trusted Advisor is a third source. AWS says that with Basic Support you can use the console for all Service Limits checks and selected checks in the Security and Fault tolerance categories, and that all checks need Business Support+, Enterprise Support or Unified Operations (AWS documentation, read 16 Nov 2026). The [options](/options/) page lists the AWS tools side by side. ## Organization-level controls beat account-by-account fixes Four of the twelve checks are set once at the organization level: [deny policies](/questions/preventive-policies/), [tamper protection](/questions/tamper-protection/), the [organization boundary](/questions/organization-boundary/) and the organization trail. Service control policies reach every member account, so a single deny covers accounts created next year. Fix the low-effort items first. [Root account MFA](/questions/root-account-mfa/), [open admin ports](/questions/no-open-admin-ports/), [IMDSv2](/questions/metadata-service-hardened/), [S3 public access](/questions/public-storage-blocked/) and [audit logging](/questions/audit-logs-everywhere/) are all rated low effort and weigh 3 in the score. ## The checklist leaves out workload and posture scanning Vulnerabilities in workloads and general posture scanning are separate jobs with their own tools and are not covered here. For an AWS-specific walk-through of a full assessment, see the [AWS security assessment](/aws-security-assessment/) guide. ## Related pages - [Azure Security Checklist for the Access Layer](/azure-security-checklist/) - [Google Cloud Security Checklist: Access Layer](/google-cloud-security-checklist/) - [Cloud Security Audit Checklist: Evidence to Collect](/cloud-security-audit-checklist/) - [All of them](/guides/) ## Sources - IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) - AWS Trusted Advisor: https://docs.aws.amazon.com/awssupport/latest/user/trusted-advisor.html (read 16 Nov 2026) - CIS Benchmarks November 2026 Update: https://www.cisecurity.org/insights/blog/cis-benchmarks-november-2026-update (read 16 Nov 2026) - IAM security best practices: https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html (read 16 Nov 2026) - Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/aws-security-checklist/ --- # Azure security checklist: 12 access checks for every tenant and subscription Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The Azure security checklist here picks 12 of the 40 access-layer checks and says where to look in Azure and Microsoft Entra ID: Conditional Access, Privileged Identity Management, network security groups, private endpoints, Azure Policy and Azure Monitor. Run them for the tenant and every subscription. Read 16 Nov 2026. ## Twelve checks cover the Azure and Entra access layer Azure access splits across two scopes: the Microsoft Entra tenant, which holds identities and Conditional Access, and the subscriptions, which hold resources and role assignments. The checks below, taken from the [assessment checklist](/cloud-security-assessment-checklist/), name the scope to open for each. | Check | Where to look in Azure | | --- | --- | | [Root or global administrator has phishing-resistant MFA and no keys](/questions/root-account-mfa/) | In Entra ID, list users holding Global Administrator and check each has a phishing-resistant authentication strength (FIDO2 or certificate) in Conditional Access. | | [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/) | List app registration client secrets and certificates with their expiry dates (Entra admin center, App registrations), and flag any valid for more than 12 months. | | [Privileged roles are granted for a limited time](/questions/no-standing-privilege/) | In Entra Privileged Identity Management, check that privileged roles show Eligible assignments with a maximum activation duration, not Active permanent. | | [Emergency accounts tested twice a year](/questions/emergency-access/) | Check that two cloud-only emergency accounts are excluded from Conditional Access policies by design and have a sign-in alert in Azure Monitor. | | [No admin or database port open to the internet](/questions/no-open-admin-ports/) | In Network security groups, search inbound rules with source Any or Internet on those ports, or use the Defender for Cloud recommendation for management ports. | | [Data services reached through private endpoints](/questions/private-service-endpoints/) | Check that Storage, SQL and Key Vault show Private endpoint connections and Public network access set to Disabled. | | [Public object storage blocked organization-wide](/questions/public-storage-blocked/) | Check that every storage account has Allow Blob anonymous access set to Disabled, and that Azure Policy denies the setting. | | [Sharing limited to your own organization](/questions/organization-boundary/) | Check tenant restrictions, B2B cross-tenant access settings and Azure Policy on storage, and review which external tenants are allowed. | | [Deny policies at the organization level](/questions/preventive-policies/) | List Azure Policy assignments with the Deny effect at management group scope. | | [Exceptions have owner, reason and expiry](/questions/exceptions-expire/) | List Azure Policy exemptions (az policy exemption list) and check each has an expiry and a category. | | [Audit logs on in every account and region](/questions/audit-logs-everywhere/) | Check Activity Log export for all subscriptions and Entra sign-in and audit log export to a Log Analytics workspace. | | [Identity alerts reach a person](/questions/identity-alerts/) | Check Azure Monitor alert rules and Entra ID Protection notifications for administrator sign-ins and role changes. | ## The Microsoft cloud security benchmark supplies the control IDs Microsoft publishes the Microsoft cloud security benchmark (MCSB) with control IDs such as IM-6 for single sign-on, PA-2 for just-in-time privileged access and NS-2 for private endpoints (Microsoft documentation, read 16 Nov 2026). The tool cites MCSB as the primary control for 26 of its 40 checks, so Azure readers can follow most checks back to a Microsoft control. Microsoft says Defender for Cloud scores recommendations that come from the MCSB into a single secure score and that a higher score means a lower identified risk level (Microsoft documentation, read 16 Nov 2026). ## Foundational CSPM is the free layer in Defender for Cloud Microsoft documents two tiers of cloud security posture management in Defender for Cloud. Foundational CSPM is available at no cost, and Defender CSPM is paid and adds attack path analysis, AI security posture and risk prioritization (Microsoft documentation, read 16 Nov 2026). The free tier is enough to answer several network checks, such as [open admin ports](/questions/no-open-admin-ports/) and [public IP addresses](/questions/no-default-public-ip/). It does not replace reading the Entra settings behind [privileged roles](/questions/no-standing-privilege/). ## Entra recommendations cover the identity checks Microsoft Entra recommendations list identity configuration improvements. Microsoft says license requirements depend on the recommendation and that some need a P2 or another license (Microsoft documentation, read 16 Nov 2026). Check the license before you plan a fix that depends on one. Microsoft also lists Privileged Identity Management as the place to make roles eligible instead of permanent, which is the fix for [standing privilege](/questions/no-standing-privilege/). The [options](/options/) page compares the Azure tools with the AWS and Google Cloud ones. ## Azure Policy turns two checks into automatic enforcement Azure Policy with the Deny effect at management group scope answers [deny policies](/questions/preventive-policies/). The same service lists [exemptions](/questions/exceptions-expire/), each of which should carry an expiry. Start new assignments in audit mode so the first report shows what a deny would have blocked. ## Related pages - [Google Cloud Security Checklist: Access Layer](/google-cloud-security-checklist/) - [Cloud Security Audit Checklist: Evidence to Collect](/cloud-security-audit-checklist/) - [Cloud Security Audit: What It Covers and Proves](/cloud-security-audit/) - [All of them](/guides/) ## Sources - Microsoft cloud security benchmark overview: https://learn.microsoft.com/en-us/security/benchmark/azure/overview (read 16 Nov 2026) - Cloud Secure Score in Defender for Cloud: https://learn.microsoft.com/en-us/azure/defender-for-cloud/secure-score-security-controls (read 16 Nov 2026) - Cloud security posture management in Defender for Cloud: https://learn.microsoft.com/en-us/azure/defender-for-cloud/concept-cloud-security-posture-management (read 16 Nov 2026) - Microsoft Entra recommendations: https://learn.microsoft.com/en-us/entra/identity/monitoring-health/overview-recommendations (read 16 Nov 2026) - Privileged Identity Management: https://learn.microsoft.com/en-us/entra/id-governance/privileged-identity-management/pim-configure (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/azure-security-checklist/ --- # Google Cloud security checklist: 12 access checks for every organization and project Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The Google Cloud security checklist here picks 12 of the 40 access-layer checks and says where to look: Cloud Identity, IAM policies, organization policy constraints, VPC Service Controls, Cloud Logging and Policy Intelligence. Run them at the organization node and in each project. Read 16 Nov 2026. ## Twelve checks cover the Google Cloud access layer Google Cloud policy attaches at the organization, folder and project levels and is inherited downward. The checks below, taken from the [assessment checklist](/cloud-security-assessment-checklist/), say which setting to read at which level. | Check | Where to look in Google Cloud | | --- | --- | | [People sign in through single sign-on with MFA](/questions/workforce-sso-mfa/) | In Cloud Identity, check that Google Cloud sign-in is bound to your SSO profile, and that IAM policies contain no consumer accounts (gmail.com). | | [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/) | Run gcloud iam service-accounts keys list for each service account with --managed-by=user, and use the Policy Intelligence service account insights to find unused keys. | | [Workloads use platform-issued identities](/questions/workload-identity/) | Check that workloads use attached service accounts or workload identity federation, and that the iam.disableServiceAccountKeyCreation org policy is enforced. | | [Administrator roles are few, named and separate](/questions/admin-roles-limited/) | Run gcloud projects get-iam-policy and the organization equivalent, and count members with roles/owner or roles/editor. | | [No public IP address by default](/questions/no-default-public-ip/) | List instances with external IPs (gcloud compute instances list) and check that the compute.vmExternalIpAccess org policy constraint is enforced. | | [Metadata service hardened](/questions/metadata-service-hardened/) | Check that GKE clusters use Workload Identity (metadata server mode GKE_METADATA) and that Compute Engine instances do not run with the default service account. | | [Public object storage blocked organization-wide](/questions/public-storage-blocked/) | Check that the storage.publicAccessPrevention org policy constraint is enforced and that buckets show Public access prevention on. | | [Sharing limited to your own organization](/questions/organization-boundary/) | Check that the iam.allowedPolicyMemberDomains org policy (domain restricted sharing) is enforced and lists only your own customer IDs. | | [Sensitive data readable from trusted networks only](/questions/trusted-network-condition/) | List VPC Service Controls perimeters and access levels, and check the sensitive projects are inside one in enforced mode, not dry run. | | [Deny policies at the organization level](/questions/preventive-policies/) | List organization policy constraints enforced at the organization node (gcloud org-policies list) and check the security-relevant ones are set. | | [Audit logs on in every account and region](/questions/audit-logs-everywhere/) | Check that Admin Activity logs are always on and that Data Access audit logs are enabled for sensitive services at the organization level. | | [Logs stored where workload admins cannot change them](/questions/logs-protected/) | Check log sinks to a bucket in a separate project with bucket lock or locked log buckets with a retention period. | ## Organization policy constraints do the work of guardrails Four of the twelve checks rest on organization policy constraints: iam.disableServiceAccountKeyCreation, compute.vmExternalIpAccess, storage.publicAccessPrevention and iam.allowedPolicyMemberDomains. Enforcing a constraint at the organization node covers every project below it, so check there before you check projects one by one. The [long-lived keys](/questions/long-lived-keys/) check is one case. Google Cloud documentation says to choose a more secure alternative for authentication and to avoid user-managed service account keys, and the [workload identity](/questions/workload-identity/) check asks whether you got there (Google Cloud documentation, read 16 Nov 2026). ## Basic roles changed in October 2026 Google Cloud IAM release notes of 29 October 2026 say the Admin, Writer and Reader basic roles are generally available and that the legacy Owner, Editor and Viewer roles remain available and maintained. Google recommends granting the new roles instead, because they support IAM Conditions (Google Cloud release notes, read 16 Nov 2026). For the [administrator roles](/questions/admin-roles-limited/) check, count both the legacy and the new basic roles. ## Security Command Center and the recommender supply evidence Google describes Security Command Center as a cloud-based risk management solution with Standard, Premium and Enterprise tiers, and its pricing page lists Standard as free of charge (Google Cloud documentation, read 16 Nov 2026). Role recommendations from the IAM recommender identify excess permissions, which feeds [access reviews](/questions/access-reviews/). The [options](/options/) page lists both beside the AWS and Azure tools. ## VPC Service Controls is the data perimeter check The [trusted network condition](/questions/trusted-network-condition/) check asks whether sensitive projects sit inside a perimeter in enforced mode. A perimeter in dry run mode logs violations but blocks nothing, so it fails the check until it is enforced. ## Related pages - [Cloud Security Audit Checklist: Evidence to Collect](/cloud-security-audit-checklist/) - [Cloud Security Audit: What It Covers and Proves](/cloud-security-audit/) - [Cloud Security Maturity Model: Five Levels](/cloud-security-maturity-model/) - [All of them](/guides/) ## Sources - Google Cloud security best practices: https://cloud.google.com/security/best-practices (read 16 Nov 2026) - IAM release notes: https://cloud.google.com/iam/docs/release-notes (read 16 Nov 2026) - Security Command Center overview: https://cloud.google.com/security-command-center/docs/concepts-security-command-center-overview (read 16 Nov 2026) - Security Command Center pricing: https://cloud.google.com/security-command-center/pricing (read 16 Nov 2026) - Best practices for managing service account keys: https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) - Role recommendations overview: https://cloud.google.com/iam/docs/recommender-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/google-cloud-security-checklist/ --- # Cloud security audit checklist: the evidence to collect for each access check Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A cloud security audit checklist pairs each access control with the evidence an auditor can inspect. This one maps the 40 checks of the assessment to artifacts such as the IAM credential report, role assignment exports and log configuration, and sets a rule for each answer: a screenshot is not evidence unless dated. ## Collect an artifact for each answer, not a statement An audit checks that a control works, so every answer on the [40-check list](/cloud-security-assessment-checklist/) needs something an outsider can inspect. A statement such as "we rotate keys" is a claim. The credential report that shows every key age is evidence. The [Word version (.docx)](/downloads/cloud-security-audit-checklist.docx) of this checklist adds three columns to each check: the artifact, the date it was produced and the person who produced it. | Domain | Evidence an auditor can ask for | Source in the cloud | | --- | --- | --- | | Identities and entitlements | Credential report with MFA and key age for each account; list of administrator assignments; last access review with removals | IAM credential report, Entra role assignments, IAM policy export | | Network reachability | Inbound rules open to the internet; list of public IP addresses; last reachability analysis with owners | Security groups, network security groups, firewall rule lists | | Data perimeter | Public access block settings; external access findings with age; list of third-party trusts with owners and end dates | S3 Block Public Access, Access Analyzer, Cloud Asset Inventory | | AI workloads | Identity per agent; written action list per agent; model allow list; invocation log sample | Role policies, model access policies, AI service logs | | Guardrails and drift | Deny policy list at the root; exception register with expiry; latest drift report | Service control policies, Azure Policy, organization policy | | Evidence | Trail configuration for every account; log retention settings; test of an alert; revocation drill record | Organization trail, log sinks, alert rules, drill notes | ## Four rules keep the evidence usable - Date every artifact. An export without a date cannot show the state on the day of the audit. - Produce it from the system, not from memory. Console exports and CLI output beat typed summaries. - Cover every account. A sample of accounts is acceptable only if the audit plan says how it was drawn. - Keep the query. Store the command or console path beside the output so the auditor can repeat it. ## Two checks are tests, not exports The [access evidence on demand](/questions/access-evidence-on-demand/) and [revocation drill](/questions/revocation-drill/) checks cannot be answered with a file. They ask whether the team can do something within a time limit, one working day for the first and one hour for the second. The evidence is a written run with the start and end time of each step. NIST SP 800-53 Rev. 5 lists the matching controls: AC-2 for account management, AU-6 for audit record review and IR-4 for incident handling. The tool cites those three for the checks above (NIST, read 16 Nov 2026). ## Audit findings should keep the assessment weights Rate each failed check with the weight it carries in the score, 1 to 3, so the audit report and the [assessment](/) agree on what matters most. The 12 checks that weigh 3 are the ones to put first in any audit summary. Track every finding with an owner and a due date in the [risk register template](/risk-register-template/). The [methodology](/methodology/) explains how the five fixes are chosen. ## Three exports start the evidence file Each cloud can produce one export that answers many checks at once. Run these first, save the output with the date and the account or subscription it came from, and then work through the rest of the list. | Cloud | Command or path | Checks it helps answer | | --- | --- | --- | | AWS | aws iam generate-credential-report, then aws iam get-credential-report | Root MFA, long-lived keys, workforce users with passwords | | Azure | az role assignment list --all, and the Entra role assignment export | Administrator roles, standing privilege, access reviews | | Google Cloud | gcloud projects get-iam-policy for each project, and the same at the organization | Administrator roles, external principals, service accounts | ## Answers marked "not known" count as failed until shown The scoring gives the lowest value to "no or not known". An audit should do the same. If nobody can produce the artifact within the audit window, record the check as failed and add the missing export to the fix list, with an owner. Record who signed off each domain. A sign-off by the owner of the control, not by the person who collected the evidence, shows that two people looked. ## This list stops at the access layer A full cloud audit also covers workload vulnerabilities, data handling and vendor contracts. Those need other checklists. For what an access-layer audit covers and how it differs from an assessment, read the [cloud security audit](/cloud-security-audit/) guide. ## Download - [Cloud security audit checklist (.docx)](/downloads/cloud-security-audit-checklist.docx): Word document made from this page, edition 16 Nov 2026 ## Related pages - [Cloud Security Audit: What It Covers and Proves](/cloud-security-audit/) - [Cloud Security Maturity Model: Five Levels](/cloud-security-maturity-model/) - [Cloud Risk Assessment: Score Gaps by Weight](/cloud-risk-assessment/) - [All of them](/guides/) ## Sources - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - IAM security best practices: https://docs.aws.amazon.com/IAM/latest/UserGuide/best-practices.html (read 16 Nov 2026) - Microsoft cloud security benchmark overview: https://learn.microsoft.com/en-us/security/benchmark/azure/overview (read 16 Nov 2026) - CSA Cloud Controls Matrix: https://cloudsecurityalliance.org/research/cloud-controls-matrix (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-audit-checklist/ --- # Cloud security audit: an access-layer audit tests controls and keeps the evidence Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A cloud security audit tests whether controls work and keeps the evidence, where an assessment scores risk. An access-layer audit covers identities, network reachability, data perimeter, AI workloads, guardrails and logs, and ends with findings that each name a control, a test and an owner. Read 16 Nov 2026. ## An audit proves control operation and an assessment estimates risk The two are often confused. An assessment asks how exposed you are and ranks the gaps. An audit asks whether a stated control operated as described over a period and keeps the proof. This site does both in sequence: the [assessment](/) gives a score in about 20 minutes, and the [audit checklist](/cloud-security-audit-checklist/) lists what to collect when someone needs to verify the answers. The guide on the [cloud security assessment](/cloud-security-assessment/) covers the first half. ## Six areas make up an access-layer audit - Identities and entitlements: who can sign in, with which factors, and what they can do. Eight checks. - Network reachability: which services answer the internet and other networks. Seven checks. - Data perimeter: who can read stored data and from which networks. Seven checks. - AI workloads: the identity, permissions, data reach and logs of AI agents. Six checks. - Guardrails and drift: policies that block risky changes and the checks that catch drift. Six checks. - Evidence: audit logs, retention, alerts and the ability to cut off access. Six checks. ## The audit steps follow a fixed order - Set the scope: list every account, subscription and project, and name the period covered. - Pick the test for each check from the [question pages](/questions/root-account-mfa/), which give AWS, Azure and Google Cloud commands. - Collect the artifact and note the date and the person who ran the test. - Rate each failed check with its weight and effort. - Write the finding with the control it fails, for example IA-2(1) in NIST SP 800-53 Rev. 5 for MFA. - Agree an owner and a due date, and recheck after the fix. ## Auditors use standards as the yardstick NIST SP 800-53 Rev. 5 is a catalog of security and privacy controls, and its families AC (access control), AU (audit and accountability), IA (identification and authentication) and IR (incident response) hold the controls behind most of the 40 checks (NIST, read 16 Nov 2026). The CIS Benchmarks give per-cloud configuration checks, and the CSA Cloud Controls Matrix gives a cloud-specific control set (CSA, read 16 Nov 2026). The tool names the source control on every check. Sites that map controls across standards are a separate subject; this guide stays on the access layer. ## A finding needs six fields Findings written in the same shape are easier to track, close and recheck. The table shows the fields and one example drawn from the [no open admin ports](/questions/no-open-admin-ports/) check. | Field | Content | Example | | --- | --- | --- | | Check | The question and its id | net-01, SSH, RDP and database ports closed to the internet | | Control | The control it fails | Google Cloud VPC firewall rules guidance; NIST SC-7 | | Test | The command or console path | List security groups with source 0.0.0.0/0 on port 22 | | Result | What the test returned, with the date | Two groups open on port 22 | | Owner | One named person | The team that owns the network account | | Due date | Date of the recheck | Set at the audit close | ## Frequent access-layer findings are old and plain Audit findings at this layer are rarely exotic. They are keys that were never rotated, administrator roles that nobody reviewed, logs kept in the account they describe and exceptions that outlived their reason. The checks for these are [long-lived keys](/questions/long-lived-keys/), [access reviews](/questions/access-reviews/), [protected logs](/questions/logs-protected/) and [expiring exceptions](/questions/exceptions-expire/). Fixing them needs a calendar and an owner more than a new tool, which is why the register records both. ## An internal audit and an external audit differ in what they accept An internal audit can accept a live demonstration. An external one usually asks for dated exports. Plan for the stricter case: run each test from a script, save the output with a timestamp and store it with the finding. The [register](/risk-register-template/) keeps the link between the finding and the artifact. ## Related pages - [Cloud Security Maturity Model: Five Levels](/cloud-security-maturity-model/) - [Cloud Risk Assessment: Score Gaps by Weight](/cloud-risk-assessment/) - [Cloud Security Assessment: Scope and Time](/cloud-security-assessment/) - [All of them](/guides/) ## Sources - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - CIS Benchmarks: https://www.cisecurity.org/cis-benchmarks (read 16 Nov 2026) - CSA Cloud Controls Matrix: https://cloudsecurityalliance.org/research/cloud-controls-matrix (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-audit/ --- # Cloud security maturity model: five levels for the cloud access layer Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The cloud security maturity model here has five levels for the access layer: Unmanaged, Reactive, Defined, Managed and Optimized. Each level is a band of the 0 to 100 domain score from the 40-question assessment and lists criteria a reviewer can observe, so two people reach the same level from the same evidence. ## Five levels map to bands of the domain score The model covers the access layer only. A team scores each of the six domains in the [assessment](/) and reads the level from the band. The bands are this site's reading convention for the 0 to 100 scale, not a published standard. | Level | Name | Domain score band | Observable criteria | | --- | --- | --- | --- | | 1 | Unmanaged | 0 to 20 | Local human users with passwords exist; admin ports are open to the internet; audit logs are on in some accounts only; nobody owns the access list. | | 2 | Reactive | 21 to 40 | Root MFA and SSO are in place; public storage is blocked in the main accounts; fixes follow incidents; no review calendar exists. | | 3 | Defined | 41 to 60 | Written standards cover roles, keys and logging; reviews happen every 90 days; organization-level deny policies exist for a few actions. | | 4 | Managed | 61 to 80 | Privileged roles are granted for limited time; drift is reported within a day; exceptions expire; alerts reach a person. | | 5 | Optimized | 81 to 100 | Revocation and evidence drills run on schedule; guardrails are tested before enforcement; AI agents have their own identity and approvals. | ## A maturity level is read per domain, not once for the account A team can reach level 4 in evidence and stay at level 2 in identities. The lowest weighted domain usually decides the real exposure, so report all six levels. The domain weights are 25, 20, 20, 15, 10 and 10, and identities and entitlements carries the largest. An overall level is optional. If you need one, compute the weighted average of the six domain scores and apply the same bands. ## Criteria are things a reviewer can see Each criterion names an artifact: a credential report, a policy list, a calendar entry, an alert rule. Words such as "mature" or "good" are not criteria. Level 3 in identities, for example, needs a written role standard and a [review every 90 days](/questions/access-reviews/) with removals recorded. Level 4 adds time-limited grants. The check is [no standing privilege](/questions/no-standing-privilege/), which weighs 3 and is rated high effort, so many teams stall between levels 3 and 4. ## How the model relates to other published models The CSA publishes the Cloud Controls Matrix, a control framework with domains and control specifications (CSA, read 16 Nov 2026). It is a list of controls, not a five-level scale. The model on this page borrows no text from it and uses its own bands. The CIS Benchmarks give configuration checks per cloud and are named on each question page. Use them as the test for a level, and use this page to turn the results into a level. ## A reviewer asks for the same artifacts at every level The reviewer asks for the artifact behind each criterion before awarding the level. For level 3 in guardrails, that is the list of organization-level deny policies. For level 4, it is a drift report with dates inside the last day, and the exception register with expiry dates. If the artifact is missing, the criterion fails and the level drops by one. This keeps the model from becoming a matter of opinion. ## The levels read differently in each domain In AI workloads, level 1 means agents run under a shared or personal credential. Level 3 means each agent has its own identity and a written action list, as in [own identity](/questions/ai-own-identity/) and [tool permissions](/questions/ai-tool-permissions/). Level 5 adds [human approval](/questions/human-approval/) for high-impact actions and logging of every call. In evidence, level 1 is audit logs on in some accounts. Level 5 is a [revocation drill](/questions/revocation-drill/) that cut off one identity within an hour and a record of the result. ## Moving up one level takes the highest weighted gaps first To go from level 2 to level 3 in a domain, fix the low-effort checks that weigh 3: [root account MFA](/questions/root-account-mfa/), [open admin ports](/questions/no-open-admin-ports/), [public storage](/questions/public-storage-blocked/) and [audit logging](/questions/audit-logs-everywhere/). The [methodology](/methodology/) explains how the five fixes are chosen, and the [options](/options/) page lists tools that report on these checks. ## Related pages - [Cloud Risk Assessment: Score Gaps by Weight](/cloud-risk-assessment/) - [Cloud Security Assessment: Scope and Time](/cloud-security-assessment/) - [AWS Security Assessment: Steps for the Access Layer](/aws-security-assessment/) - [All of them](/guides/) ## Sources - CSA Cloud Controls Matrix: https://cloudsecurityalliance.org/research/cloud-controls-matrix (read 16 Nov 2026) - CIS Benchmarks: https://www.cisecurity.org/cis-benchmarks (read 16 Nov 2026) - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-maturity-model/ --- # Cloud risk assessment: weight each gap by how much a wrong answer moves risk Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A cloud risk assessment lists the gaps in your cloud controls and ranks them by impact. This site scores the access layer with 40 questions: each answer is worth 1, 0.5 or 0, each question weighs 1 to 3, and the five fixes come from the largest weight times the missing share of the answer. ## The score is a weighted share of yes answers Each question has three answers: yes, partly and no or not known. Their values are 1, 0.5 and 0. The weight of a question runs from 1 to 3 and records how much a wrong answer moves risk. The domain score is the sum of value times weight, divided by the sum of weights, times 100. Unknown counts as no. A team that cannot show the control has the same exposure as a team without it, and the score should say so. ## A worked example scores the identities domain at 50 The eight identity questions weigh 3, 3, 3, 2, 2, 3, 2 and 1, a total of 19. Suppose the answers are yes, partly, no, partly, yes, no, partly and yes in question order. The sum of value times weight is 3 + 1.5 + 0 + 1 + 2 + 0 + 1 + 1 = 9.5, and 9.5 divided by 19 is 0.5, so the domain scores 50. | Question | Weight | Answer value | Gap (weight x (1 - value)) | | --- | --- | --- | --- | | [People sign in through single sign-on with MFA](/questions/workforce-sso-mfa/) | 3 | 0.5 | 1.5 | | [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/) | 3 | 0 | 3 | | [Workloads use platform-issued identities](/questions/workload-identity/) | 2 | 0.5 | 1 | | [Privileged roles are granted for a limited time](/questions/no-standing-privilege/) | 3 | 0 | 3 | | [Permissions reviewed every 90 days](/questions/access-reviews/) | 2 | 0.5 | 1 | ## The largest gaps become the fixes, with ties broken by effort In the example, [long-lived keys](/questions/long-lived-keys/) and [no standing privilege](/questions/no-standing-privilege/) tie at a gap of 3. Keys are rated medium effort and standing privilege high effort, so keys come first. Next comes [workforce SSO](/questions/workforce-sso-mfa/) at 1.5, then [access reviews](/questions/access-reviews/) before [workload identity](/questions/workload-identity/) at 1 each, again on lower effort. Across all six domains the tool takes the five largest gaps, breaking ties by lower effort and then by question order. ## Likelihood stays outside the arithmetic The weight captures impact if the control is missing. It does not estimate how likely an attacker is to try. Add likelihood in the [risk register template](/risk-register-template/) if your risk method needs both, and record the reason. NIST publishes a draft revision of SP 800-221, on governing ICT risk programs within enterprise risk, with comments due 7 December 2026 (NIST, read 16 Nov 2026). Teams that run an enterprise risk process can use the register to carry the fixes upward. ## Domain weights record how much of the exposure each domain holds The domain weights are chosen by this site and stated, not measured. Identities and entitlements takes 25 because standing power over the account is the common route from one stolen credential to a full breach. Network reachability and data perimeter take 20 each. AI workloads takes 15, and guardrails and evidence take 10 each. A team that disagrees can read the domain scores directly and ignore the weights. The domain score never uses the domain weight; only an overall figure does. ## Partly answers need a rule The middle answer is worth 0.5 and means the control is in place in some accounts or for some identities. Use it when the control exists but does not cover everything you own. Do not use it for plans or pilots, which count as no until they are running. Record in the register which accounts are covered, so the next run can show the gap closing. ## What a risk assessment of the access layer leaves out It does not scan workloads for vulnerabilities, check every configuration setting or rank security platforms. Those are separate jobs. For the 20-minute version of this one, see the [assessment](/), and for the full scope see the [cloud security assessment](/cloud-security-assessment/) guide. ## Related pages - [Cloud Security Assessment: Scope and Time](/cloud-security-assessment/) - [AWS Security Assessment: Steps for the Access Layer](/aws-security-assessment/) - [Cloud Risk Register Template (XLSX): Columns](/risk-register-template/) - [All of them](/guides/) ## Sources - NIST SP 800-221 Rev. 1 initial public draft: https://csrc.nist.gov/pubs/sp/800/221/r1/ipd (read 16 Nov 2026) - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - AWS Well-Architected Security Pillar: https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/welcome.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-risk-assessment/ --- # Cloud security assessment: what it covers at the access layer and how long it takes Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A cloud security assessment at the access layer scores who and what can reach your cloud accounts and data. This site's version has 40 questions in six domains, takes about 20 minutes to answer and returns a score per domain, five fixes and a PDF and XLSX export made in the browser. ## The assessment covers six domains of access The scope is the access layer: identities and entitlements, network reachability, data perimeter, AI workloads, guardrails and drift, and evidence. It matches the [40-check list](/cloud-security-assessment-checklist/). Workload vulnerabilities and posture scanning sit outside it and use other tools. ## It takes about 20 minutes to answer and longer to verify Answering the 40 questions takes about 20 minutes for a team that knows its estate. Verifying each answer against the console takes longer and depends on the number of accounts, so treat the first run as a fast screen. Each question page gives the AWS, Azure and Google Cloud steps to confirm an answer, starting with [root account MFA](/questions/root-account-mfa/). Answers stay in the browser. Nothing is sent to a server, and the PDF and XLSX exports are made in the browser. ## The output is a score, five fixes and a register - A score per domain from 0 to 100, as in [the method](/methodology/). - The five fixes with the largest weight times missing share, each with an effort rating. - A PDF for the report and an XLSX to load into a risk register, described on the [register template](/risk-register-template/) page. ## Other published self-assessments exist Wiz publishes a Cloud Security Self Assessment with 9 domains: it asks 9 introductory questions, shows a checklist for each domain answered yes and then returns a risk score (Wiz page, read 16 Nov 2026). Act offers a free access attack simulation (Act page, read 16 Nov 2026). The [options](/options/) page lists these and the provider tools, in alphabetical order and unranked. ## Prepare four things before starting - The list of accounts, subscriptions and projects, with an owner for each. - Read access to the identity provider and to the audit log destination. - One person from identity, one from network and one from data or platform, so no domain is answered by guesswork. - A place to store the exports, such as a ticket or a shared folder, with the date in the name. ## The five fixes read like a work order Each fix is one imperative sentence with an effort rating of low, medium or high. For example, the fix for [root account MFA](/questions/root-account-mfa/) is to register a hardware or passkey MFA device on every root or top administrator identity and delete any root access keys, and it is rated low effort. Low-effort fixes with a weight of 3 come first when gaps tie. The result is a short list that one team can finish in a sprint, instead of a long report that nobody owns. ## Scope choices change the verification time, not the 20 minutes The 20 minutes covers answering. Verification grows with the number of accounts and the number of clouds, because each question page lists one test per cloud. A single AWS account with one pipeline can be verified in an afternoon; an organization with hundreds of accounts needs a script for the credential report and the policy lists. Decide the scope first: the whole organization, one business unit or the accounts that hold regulated data. Write it on the first page of the export so a later run uses the same scope. ## Who reads the result The domain scores go to the security lead, who decides which domain to work on. The five fixes go to the teams that own them. The register goes to whoever tracks risk. Share the PDF with leadership only after the owners have seen their fixes, so nobody meets a finding in a meeting for the first time. ## What an assessment is not An access-layer assessment is not a vulnerability scan of workloads, a posture scan of every configuration setting or a ranking of security platforms. Each of those is a separate job. The [audit guide](/cloud-security-audit/) covers the evidence step that follows an assessment. ## Run the assessment on a schedule A single run shows a state. Two runs show a trend. Repeat after each batch of fixes, or every quarter, and keep each export. The [maturity model](/cloud-security-maturity-model/) turns the domain scores into levels you can report. A new account should first pass the [short cloud security checklist](/cloud-security-checklist/). ## Related pages - [AWS Security Assessment: Steps for the Access Layer](/aws-security-assessment/) - [Cloud Risk Register Template (XLSX): Columns](/risk-register-template/) - [Cloud Security Checklist: 15 Items for a New Account](/cloud-security-checklist/) - [All of them](/guides/) ## Sources - Wiz Cloud Security Self Assessment: https://www.wiz.io/cloud-security-assessment (read 16 Nov 2026) - Act free assessment: https://act.security/free-assessment (read 16 Nov 2026) - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - AWS Well-Architected Security Pillar: https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/welcome.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-assessment/ --- # AWS security assessment: run the access-layer steps across the whole organization Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. An AWS security assessment of the access layer starts at the organization root: list the accounts, pull the IAM credential report, read the service control policies, turn on IAM Access Analyzer and check the organization trail. This guide gives the order, the AWS service behind each step and the check it answers. ## Seven steps cover an AWS organization - List every account in AWS Organizations, including suspended and recently created ones. - Pull the IAM credential report for each account and read root MFA and key age. - Read the service control policies at the root and in each organizational unit. - Create an IAM Access Analyzer for the organization and read the external access findings. - Check security groups and public IP addresses for open admin and database ports. - Check S3 Block Public Access, data perimeter conditions and customer-managed key policies. - Check the organization trail, the log archive account and alerts for root sign-in and policy changes. ## Each step maps to checks on the 40-check list Step 2 answers [root account MFA](/questions/root-account-mfa/) and [long-lived keys](/questions/long-lived-keys/). Step 3 answers [preventive policies](/questions/preventive-policies/) and [tamper protection](/questions/tamper-protection/). Step 4 answers [external access findings](/questions/external-access-findings/). Step 7 answers [audit logs](/questions/audit-logs-everywhere/) and [protected logs](/questions/logs-protected/). The AWS-specific commands for these and more are on the [AWS security checklist](/aws-security-checklist/). ## AWS tools report on parts of the work AWS says IAM Access Analyzer provides external access, internal access and unused access analyzers, plus policy validation and policy generation from CloudTrail activity (AWS documentation, read 16 Nov 2026). It charges for unused access analysis by the number of roles and users analyzed (AWS documentation, read 16 Nov 2026). AWS Security Hub CSPM checks controls against standards such as AWS Foundational Security Best Practices and CIS, and AWS says new accounts get a 30-day free trial (AWS documentation, read 16 Nov 2026). The [options](/options/) page lists both with their makers' descriptions. ## The AWS Well-Architected Security Pillar is the provider yardstick AWS publishes the Security Pillar of the AWS Well-Architected Framework as its guidance on securing workloads (AWS documentation, read 16 Nov 2026). The tool cites AWS documentation as the primary control for 7 of its 40 checks, where no CIS or NIST control fits the check directly. Combine this with the scored [assessment](/) to get the order of fixes. ## Command for each step The commands below are the AWS CLI forms of the seven steps. Run them with credentials that have read-only access in each account. | Step | Command or console path | | --- | --- | | List accounts | aws organizations list-accounts | | Credential report | aws iam generate-credential-report, then aws iam get-credential-report | | Service control policies | aws organizations list-policies --filter SERVICE_CONTROL_POLICY | | External access analyzer | aws accessanalyzer create-analyzer --type ORGANIZATION --analyzer-name org-external | | Open ports | aws ec2 describe-security-groups --filters Name=ip-permission.cidr,Values=0.0.0.0/0 | | Public access block | aws s3control get-public-access-block --account-id ACCOUNT_ID | | Trails | aws cloudtrail describe-trails --include-shadow-trails | ## Cross-account roles deserve their own pass List roles whose trust policy names an external account and check for an external ID or organization condition, as in [third-party access](/questions/third-party-access/). AWS describes IAM Access Analyzer external access findings as covering resources such as S3 buckets and IAM roles shared with an external entity (AWS documentation, read 16 Nov 2026). Review those findings weekly, as the [external access findings](/questions/external-access-findings/) check asks. Add AI services to the pass. If Amazon Bedrock agents run in the account, check that each has its own role, as in [own identity](/questions/ai-own-identity/). ## Do the work from the management account only where needed Organization-level reads, such as the list of service control policies, run from the management account or a delegated administrator. Everything else should run from a read-only role assumed into each member account, so the assessment itself follows the same rule as the [administrator roles](/questions/admin-roles-limited/) check and leaves a trail in CloudTrail. ## An AWS-only assessment misses cross-cloud trust If the organization also runs Azure or Google Cloud, a service account or app registration can hold keys into AWS. The [workload identity](/questions/workload-identity/) check asks whether workloads use platform-issued identities, which removes most of those secrets. Run the full [cloud security assessment](/cloud-security-assessment/) if you use more than one cloud. ## Related pages - [Cloud Risk Register Template (XLSX): Columns](/risk-register-template/) - [Cloud Security Checklist: 15 Items for a New Account](/cloud-security-checklist/) - [Cloud Security Assessment Checklist: 40 Checks](/cloud-security-assessment-checklist/) - [All of them](/guides/) ## Sources - IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) - AWS Security Hub CSPM: https://docs.aws.amazon.com/securityhub/latest/userguide/what-is-securityhub.html (read 16 Nov 2026) - AWS Well-Architected Security Pillar: https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/welcome.html (read 16 Nov 2026) - Service control policies: https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/aws-security-assessment/ --- # Cloud risk register template: 15 columns, a gap score and a named owner Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The cloud risk register template is an XLSX made in your browser from the assessment answers. It has 15 columns, from risk ID and check ID to gap, owner, due date and evidence link. The assessment fills 8 and calculates one; you add the risk ID, owner, due date, status, evidence link and review date. ## The register has 15 columns, 8 of them filled from your answers After you finish the [assessment](/), the export button builds the XLSX in the browser. Nothing is uploaded. Each answer that scores below yes becomes a row, so a team with 12 gaps gets 12 rows. The columns are listed below. The risk ID, owner, due date, status, evidence link and review date columns are blank on purpose, because only your team can fill them. | Column | What goes in it | Filled from | | --- | --- | --- | | Risk ID | A short stable id such as R-001 | You | | Check ID | The question id, for example id-03 | Assessment | | Domain | One of the six domains | Assessment | | Risk statement | The gap in one sentence, written as the question reversed | Assessment | | Answer | Yes, partly or no or not known, with its value 1, 0.5 or 0 | Assessment | | Weight | 1 to 3, from the question | Assessment | | Gap | Weight times (1 minus value) | Calculated | | Fix | The imperative sentence from the question page | Assessment | | Effort | Low, medium or high | Assessment | | Control | The framework, reference and short title cited by the question | Assessment | | Owner | One named person or team | You | | Due date | The date of the recheck | You | | Status | Open, in progress, accepted or closed | You | | Evidence link | Where the dated artifact is stored | You | | Review date | The next date the answer will be tested again | You | ## The gap column ranks the rows Gap is weight times one minus the answer value. A no on a weight 3 question is a gap of 3, and a partly on a weight 2 question is a gap of 1. Sort the sheet by gap, descending, and the top rows are the fixes the assessment would give. The tool shows five; the register keeps all of them. Where two rows tie, order them by effort, low first, then by question order, the same rule as in the [methodology](/methodology/). ## Every row needs one owner and one date A register without owners is a list of worries. Put one named person on each row, a team name if the person is not known yet, and a due date that is a recheck of the test, not a hope. Link the dated artifact in the evidence column, such as the credential report that shows a key was deleted. Rows with a status of accepted need a reason and an end date. That follows the [exceptions expire](/questions/exceptions-expire/) check: an accepted risk with no end date is an exception with no expiry. ## Add likelihood only if your method needs it Many risk methods score impact and likelihood. The register carries impact in the weight and the gap, and leaves likelihood out. If your process needs it, add a column and say how you set it, then keep the assessment gap as the first sort key so the order stays reproducible. The [cloud risk assessment](/cloud-risk-assessment/) guide shows a worked score. The register is how that result becomes tracked work. ## One filled row shows the shape Row R-001 could read: check id-03, domain identities, risk statement "Long-lived access keys exist and are older than 90 days", answer no (0), weight 3, gap 3, fix "Replace static keys with short-lived credentials, and delete every key older than 90 days or unused for 90 days", effort medium, owner the platform team, due date the next recheck, status in progress. The [long-lived keys](/questions/long-lived-keys/) page gives the AWS, Azure and Google Cloud test for the evidence link. ## Update the register on every run Run the assessment again after the fixes. Rows whose answer moved to yes are closed with the date; new gaps are added; rows that did not move keep their owners and get a new review date. Keep each export, since two exports show a trend that one cannot. The [maturity model](/cloud-security-maturity-model/) turns the domain scores into levels for reporting, and the [options](/options/) page lists tools that report on some of the same checks. ## Related pages - [Cloud Security Checklist: 15 Items for a New Account](/cloud-security-checklist/) - [Cloud Security Assessment Checklist: 40 Checks](/cloud-security-assessment-checklist/) - [AWS Security Checklist for the Access Layer](/aws-security-checklist/) - [All of them](/guides/) ## Sources - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) - NIST SP 800-221 Rev. 1 initial public draft: https://csrc.nist.gov/pubs/sp/800/221/r1/ipd (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/risk-register-template/ --- # Cloud security checklist: 15 items to finish before a new account takes traffic Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The cloud security checklist here has 15 items for a new AWS, Azure or Google Cloud account, taken from the 40 assessment checks: five on identity, two on network, two on data, two on guardrails and four on logs and alerts. Finish it on day one, before the account takes production traffic. ## Fifteen items are enough for day one A new account needs a small list that one person can finish in a day. The 15 items below are a subset of the 40 checks in the [assessment checklist](/cloud-security-assessment-checklist/), chosen because each is low or medium effort when the account is empty and expensive to retrofit later. - [Root or global administrator has phishing-resistant MFA and no keys](/questions/root-account-mfa/): Turn on phishing-resistant MFA for the root or top administrator and delete any root keys. - [People sign in through single sign-on with MFA](/questions/workforce-sso-mfa/): Connect people to the cloud through single sign-on and disable local users. - [Long-lived keys removed or rotated every 90 days](/questions/long-lived-keys/): Create no long-lived access keys; if one exists, set a 90-day rotation. - [Administrator roles are few, named and separate](/questions/admin-roles-limited/): Name the few administrators and give each a separate daily account. - [Emergency accounts tested twice a year](/questions/emergency-access/): Create two emergency accounts and alert on every sign-in. - [No admin or database port open to the internet](/questions/no-open-admin-ports/): Close SSH, RDP and database ports to the whole internet. - [No public IP address by default](/questions/no-default-public-ip/): Deny public IP addresses by default on compute and databases. - [Public object storage blocked organization-wide](/questions/public-storage-blocked/): Block public access to object storage at the organization level. - [External access findings reviewed weekly](/questions/external-access-findings/): Turn on external access analysis and send findings to an owner. - [Deny policies at the organization level](/questions/preventive-policies/): Write organization-level deny policies for your riskiest actions. - [Guardrails protected from tampering](/questions/tamper-protection/): Stop everyone but a security role from turning off logs and guardrails. - [Audit logs on in every account and region](/questions/audit-logs-everywhere/): Turn on audit logs in every region of the account. - [Logs stored where workload admins cannot change them](/questions/logs-protected/): Send logs to a separate security account and lock them. - [Logs kept 365 days, 90 searchable](/questions/log-retention/): Keep audit logs for at least 365 days. - [Identity alerts reach a person](/questions/identity-alerts/): Alert on root sign-in, policy changes, new keys and disabled logging. ## The order matters more than the list Do identity first, because every later setting depends on who can change it. Then close the network, then block public data, then write the deny policies, and turn on logs before anyone deploys. Logs switched on later cannot show what happened earlier. If you do only three things, do [root account MFA](/questions/root-account-mfa/), [audit logging](/questions/audit-logs-everywhere/) and [public storage blocking](/questions/public-storage-blocked/). Each is rated low effort and weighs 3 in the score. ## Where the list stops The list leaves out the AI workload checks and several data perimeter checks, because a new account rarely has either on day one. Add them when the first agent or sensitive data store arrives. The [AI identity](/questions/ai-own-identity/) and [sensitive data location](/questions/sensitive-data-located/) checks are the first to add. The list is a floor. It does not tell you the state of an older estate, which needs the scored [assessment](/). ## Use the provider guidance for settings the list does not cover The three providers publish their own baselines: the AWS Well-Architected Security Pillar, the Microsoft cloud security benchmark and Google Cloud security best practices. All three are linked below and each question page cites the one that applies (read 16 Nov 2026). Run the 15 items again after the first month, then run the full [40-check list](/cloud-security-assessment-checklist/). The [options](/options/) page lists tools that can report on several of these items for you. ## Three tests, one per cloud For [no open admin ports](/questions/no-open-admin-ports/), AWS shows it in security group rules with source 0.0.0.0/0, Azure in network security group inbound rules with source Any or Internet, and Google Cloud in firewall rules with source range 0.0.0.0/0. The question pages give the same split for each of the 15 items, so a team on any of the three clouds can run the list. ## Record the result for the account Write the date, the person and the evidence for each item, and store them with the account record. A one-page record lets a later audit see that the account started clean, and it gives the [risk register template](/risk-register-template/) a baseline if anything regresses. ## Related pages - [Cloud Security Assessment Checklist: 40 Checks](/cloud-security-assessment-checklist/) - [AWS Security Checklist for the Access Layer](/aws-security-checklist/) - [Azure Security Checklist for the Access Layer](/azure-security-checklist/) - [All of them](/guides/) ## Sources - AWS Well-Architected Security Pillar: https://docs.aws.amazon.com/wellarchitected/latest/security-pillar/welcome.html (read 16 Nov 2026) - Microsoft cloud security benchmark overview: https://learn.microsoft.com/en-us/security/benchmark/azure/overview (read 16 Nov 2026) - Google Cloud security best practices: https://cloud.google.com/security/best-practices (read 16 Nov 2026) - NIST SP 800-53 Rev. 5: https://csrc.nist.gov/pubs/sp/800/53/r5/upd1/final (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/cloud-security-checklist/ --- # 30 terms, each with a sourced example Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Plain definitions of 30 terms used in the assessment, from access analyzer and break-glass account to service control policy and workload identity federation. Each definition stands alone, cites a provider or NIST page that was read, shows the read date and names the check it supports. - [Access Analyzer](/glossary/access-analyzer/): IAM Access Analyzer is an AWS service that analyzes policies to find resources shared with outside principals, access that is never used, and access paths to chosen resources. - [Audit log](/glossary/audit-log/): An audit log is a time-ordered record of who did what in a cloud account, at what time and from where. - [Break-glass account](/glossary/break-glass-account/): A break-glass account is an emergency administrator account kept for the case where normal sign-in fails, for example when an identity provider is down. - [Brokered access](/glossary/brokered-access/): Brokered access means administrators reach a server through a managed, logged service instead of an open inbound port. - [Conditional Access](/glossary/conditional-access/): Conditional Access is Microsoft Entra's policy engine that decides whether to allow a sign-in, based on signals such as user, device, location and risk. - [Data perimeter](/glossary/data-perimeter/): A data perimeter is a set of preventive controls that keep trusted identities, accessing from expected networks, using trusted resources. - [Deny policy](/glossary/deny-policy/): A deny policy is an access rule that blocks a principal from an action whatever allow rules say, and in Google Cloud it is evaluated before allow policies. - [Drift](/glossary/drift/): Drift is the difference between the configuration a team intended and the configuration a cloud account actually holds, usually from manual changes or exceptions. - [Guardrail](/glossary/guardrail/): A guardrail is a rule that limits what actions are possible in a cloud account, set above the level where workload teams work. - [IAM condition](/glossary/iam-condition/): An IAM condition is an attribute test attached to a role binding or policy so the grant applies only when the test is true, such as before a given time or from a given resource. - [IMDSv2](/glossary/imdsv2/): IMDSv2 is the session-oriented version of the AWS EC2 instance metadata service. - [Just-in-time access](/glossary/just-in-time-access/): Just-in-time access grants a privileged role or network port only when a person asks, for a limited time, and takes it away afterward. - [Least privilege](/glossary/least-privilege/): NIST defines least privilege as the principle that a security architecture is designed so that each entity is granted the minimum system resources and authorizations that the entity needs to perform its function. - [Managed identity](/glossary/managed-identity/): A managed identity is an identity that Azure creates and manages for a resource, so code can obtain tokens for other services without a stored password or key. - [Multifactor authentication](/glossary/multifactor-authentication/): Multifactor authentication, or MFA, requires a second proof of identity beyond a password, such as a code, an app prompt or a security key. - [Organization policy](/glossary/organization-policy/): An organization policy is a Google Cloud setting that restricts how resources can be configured across the organization, folders and projects, through named constraints. - [Phishing-resistant MFA](/glossary/phishing-resistant-mfa/): Phishing-resistant MFA is authentication that cannot be replayed by a fake sign-in page, because the credential is bound to the real site or device. - [Policy Analyzer](/glossary/policy-analyzer/): Policy Analyzer is a Google Cloud Policy Intelligence tool that answers who can access a given resource and how, by analyzing IAM allow policies. - [Principal access boundary](/glossary/principal-access-boundary/): A principal access boundary is a Google Cloud policy that sets the most resources a group of principals can ever be given access to, whatever roles they hold. - [Privileged Identity Management](/glossary/privileged-identity-management/): Privileged Identity Management, or PIM, is a Microsoft Entra service that lets organizations give just-in-time and time-bound access to Entra and Azure roles, require approval and justification, and review assignments. - [Public access prevention](/glossary/public-access-prevention/): Public access prevention is a Google Cloud Storage setting that blocks a bucket or all buckets in a project or organization from being made public, whatever the permissions say. - [Resource lock](/glossary/resource-lock/): A resource lock is an Azure setting that stops a resource from being deleted or changed, whoever holds the role to do so. - [Root user](/glossary/root-user/): The AWS account root user is the identity created with the account, with complete access to all services and resources in it. - [Secure score](/glossary/secure-score/): Secure score is Microsoft Defender for Cloud's single number that aggregates security findings, where a higher score means a lower identified risk level. - [Separation of duty](/glossary/separation-of-duty/): NIST defines separation of duty as the principle that no user should be given enough privileges to misuse the system on their own. - [Service account key](/glossary/service-account-key/): A service account key is a long-lived credential for a Google Cloud service account, held as a file. - [Service control policy](/glossary/service-control-policy/): A service control policy, or SCP, is an AWS Organizations policy that sets the maximum permissions for accounts in an organization or organizational unit. - [Standing privilege](/glossary/standing-privilege/): Standing privilege is access that a person or workload holds all the time, whether or not it is in use. - [VPC Service Controls](/glossary/vpc-service-controls/): VPC Service Controls is a Google Cloud feature that creates a perimeter around supported services and projects, to reduce the risk of data exfiltration by blocking requests that cross it. - [Workload identity federation](/glossary/workload-identity-federation/): Workload identity federation lets a workload outside a cloud, such as a CI pipeline or another cloud, obtain short-lived credentials by presenting a token from a trusted identity provider, instead of holding a stored key. ## Related pages - [News](/news/) - [The 40 questions of the assessment](/questions/) - [Cloud security assessment tools and options](/options/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/ --- # Access Analyzer Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. IAM Access Analyzer is an AWS service that analyzes policies to find resources shared with outside principals, access that is never used, and access paths to chosen resources. It produces findings that an owner reviews and closes, and it also validates policies and generates them from CloudTrail activity. ## An example, sourced The external access check asks that someone reviews these findings weekly. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Audit log: meaning and an example](/glossary/audit-log/) - [Break-glass account: meaning and an example](/glossary/break-glass-account/) - [Brokered access: meaning and an example](/glossary/brokered-access/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/access-analyzer/ --- # Audit log Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. An audit log is a time-ordered record of who did what in a cloud account, at what time and from where. In Google Cloud the main kinds are Admin Activity and Data Access logs. Audit logs are the evidence used to detect misuse, investigate incidents and prove which identity changed a setting. ## An example, sourced The evidence checks ask that audit logs are on in every account and region and stored where workload administrators cannot change them. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Break-glass account: meaning and an example](/glossary/break-glass-account/) - [Brokered access: meaning and an example](/glossary/brokered-access/) - [Conditional Access: meaning and an example](/glossary/conditional-access/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/logging/docs/audit (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/audit-log/ --- # Break-glass account Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A break-glass account is an emergency administrator account kept for the case where normal sign-in fails, for example when an identity provider is down. Microsoft calls them emergency access accounts. They hold long unique credentials, are exempt from the usual sign-in policies by design, and trigger an alert on every use. ## An example, sourced The emergency access check asks that two such accounts exist, are monitored and are tested at least twice a year. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Brokered access: meaning and an example](/glossary/brokered-access/) - [Conditional Access: meaning and an example](/glossary/conditional-access/) - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/identity/role-based-access-control/security-emergency-access (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/break-glass-account/ --- # Brokered access Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Brokered access means administrators reach a server through a managed, logged service instead of an open inbound port. Azure Bastion is Microsoft's example, and AWS Systems Manager Session Manager and Google Cloud Identity-Aware Proxy fill the same role. The broker checks identity, records the session and keeps SSH and RDP closed to the internet. ## An example, sourced The brokered admin access check asks that no inbound admin ports are needed. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Conditional Access: meaning and an example](/glossary/conditional-access/) - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [Deny policy: meaning and an example](/glossary/deny-policy/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/bastion/bastion-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/brokered-access/ --- # Conditional Access Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Conditional Access is Microsoft Entra's policy engine that decides whether to allow a sign-in, based on signals such as user, device, location and risk. Administrators write policies that can require multifactor authentication or block access. It is the usual place to enforce MFA for every user and every cloud app. ## An example, sourced The workforce sign-in check looks for a policy that requires MFA for all users and all cloud apps. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Data perimeter: meaning and an example](/glossary/data-perimeter/) - [Deny policy: meaning and an example](/glossary/deny-policy/) - [Drift: meaning and an example](/glossary/drift/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/identity/conditional-access/overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/conditional-access/ --- # Data perimeter Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A data perimeter is a set of preventive controls that keep trusted identities, accessing from expected networks, using trusted resources. AWS describes three objectives: only trusted identities, only trusted resources and only expected networks. A perimeter limits damage when one policy or credential is wrong, because a second control still blocks the request. ## An example, sourced The data perimeter domain holds seven checks, from public storage blocking to third-party access. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Deny policy: meaning and an example](/glossary/deny-policy/) - [Drift: meaning and an example](/glossary/drift/) - [Guardrail: meaning and an example](/glossary/guardrail/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/whitepapers/latest/building-a-data-perimeter-on-aws/building-a-data-perimeter-on-aws.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/data-perimeter/ --- # Deny policy Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A deny policy is an access rule that blocks a principal from an action whatever allow rules say, and in Google Cloud it is evaluated before allow policies. Deny rules are the main way to keep a mistaken grant from taking effect, and they are used for guardrails that no administrator in a workload account can lift. ## An example, sourced The tamper protection check asks that logging and security services are protected by rules of this kind. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Drift: meaning and an example](/glossary/drift/) - [Guardrail: meaning and an example](/glossary/guardrail/) - [IAM condition: meaning and an example](/glossary/iam-condition/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/deny-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/deny-policy/ --- # Drift Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Drift is the difference between the configuration a team intended and the configuration a cloud account actually holds, usually from manual changes or exceptions. A drift check compares live settings with rules and reports each difference. AWS Config is one service that records configuration and evaluates it against rules. ## An example, sourced The drift detection check asks that differences are reported within a day. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Guardrail: meaning and an example](/glossary/guardrail/) - [IAM condition: meaning and an example](/glossary/iam-condition/) - [IMDSv2: meaning and an example](/glossary/imdsv2/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/config/latest/developerguide/WhatIsConfig.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/drift/ --- # Guardrail Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A guardrail is a rule that limits what actions are possible in a cloud account, set above the level where workload teams work. AWS service control policies are one kind. Guardrails can be preventive, blocking the action, or detective, reporting it afterward. They protect an organization from mistakes by people who hold broad permissions. ## An example, sourced The guardrails and drift domain holds six checks, from deny policies to exception expiry. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [IAM condition: meaning and an example](/glossary/iam-condition/) - [IMDSv2: meaning and an example](/glossary/imdsv2/) - [Just-in-time access: meaning and an example](/glossary/just-in-time-access/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/guardrail/ --- # IAM condition Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. An IAM condition is an attribute test attached to a role binding or policy so the grant applies only when the test is true, such as before a given time or from a given resource. In Google Cloud, IAM Conditions are an expression on a role binding. Conditions narrow a broad role to a specific time, resource or request. ## An example, sourced The no standing privilege check accepts time-based conditions on privileged roles as one way to limit duration. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [IMDSv2: meaning and an example](/glossary/imdsv2/) - [Just-in-time access: meaning and an example](/glossary/just-in-time-access/) - [Least privilege: meaning and an example](/glossary/least-privilege/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/conditions-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/iam-condition/ --- # IMDSv2 Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. IMDSv2 is the session-oriented version of the AWS EC2 instance metadata service. It requires a token obtained with a PUT request before any metadata is returned, which blocks many request-forgery attacks that read credentials from the metadata address. Instances can be set to require IMDSv2 and reject version 1. ## An example, sourced The metadata service check asks that IMDSv2 is required on all instances. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Just-in-time access: meaning and an example](/glossary/just-in-time-access/) - [Least privilege: meaning and an example](/glossary/least-privilege/) - [Managed identity: meaning and an example](/glossary/managed-identity/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/configuring-instance-metadata-service.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/imdsv2/ --- # Just-in-time access Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Just-in-time access grants a privileged role or network port only when a person asks, for a limited time, and takes it away afterward. Microsoft Defender for Cloud offers it for virtual machine ports and Privileged Identity Management offers it for roles. The goal is that no administrator holds standing power between tasks. ## An example, sourced The no standing privilege check asks that privileged roles are granted this way. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Least privilege: meaning and an example](/glossary/least-privilege/) - [Managed identity: meaning and an example](/glossary/managed-identity/) - [Multifactor authentication: meaning and an example](/glossary/multifactor-authentication/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/defender-for-cloud/just-in-time-access-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/just-in-time-access/ --- # Least privilege Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. NIST defines least privilege as the principle that a security architecture is designed so that each entity is granted the minimum system resources and authorizations that the entity needs to perform its function. In cloud, that means each identity gets only the actions, resources and conditions its task needs. ## An example, sourced Several identity checks, including administrator roles and access reviews, test whether this holds in practice. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Managed identity: meaning and an example](/glossary/managed-identity/) - [Multifactor authentication: meaning and an example](/glossary/multifactor-authentication/) - [Organization policy: meaning and an example](/glossary/organization-policy/) - [All of them](/glossary/) ## Sources - https://csrc.nist.gov/glossary/term/least_privilege (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/least-privilege/ --- # Managed identity Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A managed identity is an identity that Azure creates and manages for a resource, so code can obtain tokens for other services without a stored password or key. System-assigned identities live with one resource and user-assigned identities can be shared. Managed identities remove secrets that would otherwise sit in code or settings. ## An example, sourced The workload identity check asks that apps use managed identities instead of client secrets. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Multifactor authentication: meaning and an example](/glossary/multifactor-authentication/) - [Organization policy: meaning and an example](/glossary/organization-policy/) - [Phishing-resistant MFA: meaning and an example](/glossary/phishing-resistant-mfa/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/identity/managed-identities-azure-resources/overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/managed-identity/ --- # Multifactor authentication Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Multifactor authentication, or MFA, requires a second proof of identity beyond a password, such as a code, an app prompt or a security key. Microsoft documents several methods and their strength. MFA lowers the chance that a stolen password alone gives access, but methods differ in how well they resist phishing. ## An example, sourced The root account MFA and workforce sign-in checks both depend on it. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Organization policy: meaning and an example](/glossary/organization-policy/) - [Phishing-resistant MFA: meaning and an example](/glossary/phishing-resistant-mfa/) - [Policy Analyzer: meaning and an example](/glossary/policy-analyzer/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/identity/authentication/concept-mfa-howitworks (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/multifactor-authentication/ --- # Organization policy Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. An organization policy is a Google Cloud setting that restricts how resources can be configured across the organization, folders and projects, through named constraints. Examples include disabling service account key creation or restricting external IP addresses. Policies are inherited downward, so one setting at the organization node covers every project below. ## An example, sourced Several checks, such as public storage blocking and external IP restrictions, rest on named constraints. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Phishing-resistant MFA: meaning and an example](/glossary/phishing-resistant-mfa/) - [Policy Analyzer: meaning and an example](/glossary/policy-analyzer/) - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/resource-manager/docs/organization-policy/overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/organization-policy/ --- # Phishing-resistant MFA Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Phishing-resistant MFA is authentication that cannot be replayed by a fake sign-in page, because the credential is bound to the real site or device. Microsoft lists passkeys and FIDO2 security keys, and certificate-based authentication, as examples. Codes sent by text message or typed from an app can still be phished. ## An example, sourced The root account check asks for phishing-resistant MFA, not only any second factor. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Policy Analyzer: meaning and an example](/glossary/policy-analyzer/) - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [Privileged Identity Management: meaning and an example](/glossary/privileged-identity-management/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/identity/authentication/concept-authentication-passkeys-fido2 (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/phishing-resistant-mfa/ --- # Policy Analyzer Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Policy Analyzer is a Google Cloud Policy Intelligence tool that answers who can access a given resource and how, by analyzing IAM allow policies. Teams use it to prepare access evidence and to check the effect of a change before it is made. Results can be exported for an auditor. ## An example, sourced The access evidence check suggests running a query like this for one sensitive data store. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Principal access boundary: meaning and an example](/glossary/principal-access-boundary/) - [Privileged Identity Management: meaning and an example](/glossary/privileged-identity-management/) - [Public access prevention: meaning and an example](/glossary/public-access-prevention/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/policy-intelligence/docs/policy-analyzer-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/policy-analyzer/ --- # Principal access boundary Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A principal access boundary is a Google Cloud policy that sets the most resources a group of principals can ever be given access to, whatever roles they hold. It limits the reach of an identity from above, so a mistaken role binding cannot extend access beyond the boundary. ## An example, sourced The organization boundary check looks for controls of this type that limit sharing and reach. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Privileged Identity Management: meaning and an example](/glossary/privileged-identity-management/) - [Public access prevention: meaning and an example](/glossary/public-access-prevention/) - [Resource lock: meaning and an example](/glossary/resource-lock/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/principal-access-boundary-policies (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/principal-access-boundary/ --- # Privileged Identity Management Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Privileged Identity Management, or PIM, is a Microsoft Entra service that lets organizations give just-in-time and time-bound access to Entra and Azure roles, require approval and justification, and review assignments. It replaces permanent role assignments with eligible ones that a person activates when needed. ## An example, sourced The no standing privilege check looks for eligible assignments in PIM with a maximum activation time. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Public access prevention: meaning and an example](/glossary/public-access-prevention/) - [Resource lock: meaning and an example](/glossary/resource-lock/) - [Root user: meaning and an example](/glossary/root-user/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/entra/id-governance/privileged-identity-management/pim-configure (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/privileged-identity-management/ --- # Public access prevention Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Public access prevention is a Google Cloud Storage setting that blocks a bucket or all buckets in a project or organization from being made public, whatever the permissions say. AWS offers S3 Block Public Access and Azure offers an account setting for anonymous blob access with the same aim. ## An example, sourced The public storage check asks that the block is set at the organization level. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Resource lock: meaning and an example](/glossary/resource-lock/) - [Root user: meaning and an example](/glossary/root-user/) - [Secure score: meaning and an example](/glossary/secure-score/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/storage/docs/public-access-prevention (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/public-access-prevention/ --- # Resource lock Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A resource lock is an Azure setting that stops a resource from being deleted or changed, whoever holds the role to do so. The two lock levels are delete and read-only. Locks are used to protect log storage and diagnostic settings from tampering by workload administrators. ## An example, sourced The tamper protection check mentions locks as one way to protect logging. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Root user: meaning and an example](/glossary/root-user/) - [Secure score: meaning and an example](/glossary/secure-score/) - [Separation of duty: meaning and an example](/glossary/separation-of-duty/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/azure-resource-manager/management/lock-resources (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/resource-lock/ --- # Root user Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The AWS account root user is the identity created with the account, with complete access to all services and resources in it. AWS recommends not using it for daily tasks and protecting it with MFA. Azure and Google Cloud have the equivalents in Global Administrator and the super administrator. ## An example, sourced The first check of the assessment asks that the root user has phishing-resistant MFA and no access keys. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Secure score: meaning and an example](/glossary/secure-score/) - [Separation of duty: meaning and an example](/glossary/separation-of-duty/) - [Service account key: meaning and an example](/glossary/service-account-key/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/IAM/latest/UserGuide/id_root-user.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/root-user/ --- # Secure score Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Secure score is Microsoft Defender for Cloud's single number that aggregates security findings, where a higher score means a lower identified risk level. It is based on built-in recommendations from the Microsoft cloud security benchmark. It is a posture measure across many settings, and it is not the same as this site's access-layer score. ## An example, sourced The assessment's domain scores use only the 40 access-layer checks. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Separation of duty: meaning and an example](/glossary/separation-of-duty/) - [Service account key: meaning and an example](/glossary/service-account-key/) - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [All of them](/glossary/) ## Sources - https://learn.microsoft.com/en-us/azure/defender-for-cloud/secure-score-security-controls (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/secure-score/ --- # Separation of duty Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. NIST defines separation of duty as the principle that no user should be given enough privileges to misuse the system on their own. In cloud, it means the people who change a control are not the only people who can approve it or erase its logs. It is a reason for separate security and workload roles. ## An example, sourced The log protection check applies it to audit logs. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Service account key: meaning and an example](/glossary/service-account-key/) - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [Standing privilege: meaning and an example](/glossary/standing-privilege/) - [All of them](/glossary/) ## Sources - https://csrc.nist.gov/glossary/term/separation_of_duty (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/separation-of-duty/ --- # Service account key Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A service account key is a long-lived credential for a Google Cloud service account, held as a file. Google says to choose more secure alternatives and to avoid user-managed keys, because a key works wherever it is copied until it is deleted. AWS access keys and Azure client secrets present the same risk. ## An example, sourced The long-lived keys check asks that keys are absent or rotated every 90 days. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Service control policy: meaning and an example](/glossary/service-control-policy/) - [Standing privilege: meaning and an example](/glossary/standing-privilege/) - [VPC Service Controls: meaning and an example](/glossary/vpc-service-controls/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/service-account-key/ --- # Service control policy Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. A service control policy, or SCP, is an AWS Organizations policy that sets the maximum permissions for accounts in an organization or organizational unit. It does not grant access. It filters what member-account roles can do, including the account's own administrators, which makes it the main guardrail in AWS. ## An example, sourced The preventive policies check asks for SCPs that deny the riskiest actions. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Standing privilege: meaning and an example](/glossary/standing-privilege/) - [VPC Service Controls: meaning and an example](/glossary/vpc-service-controls/) - [Workload identity federation: meaning and an example](/glossary/workload-identity-federation/) - [All of them](/glossary/) ## Sources - https://docs.aws.amazon.com/organizations/latest/userguide/orgs_manage_policies_scps.html (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/service-control-policy/ --- # Standing privilege Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Standing privilege is access that a person or workload holds all the time, whether or not it is in use. Google Cloud Privileged Access Manager grants temporary elevated access instead. Standing administrator roles expand the damage from a stolen credential, because the power is already there when the credential is taken. ## An example, sourced Weighted 3 in the score, the no standing privilege check tests for its absence. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [VPC Service Controls: meaning and an example](/glossary/vpc-service-controls/) - [Workload identity federation: meaning and an example](/glossary/workload-identity-federation/) - [Access Analyzer: meaning and an example](/glossary/access-analyzer/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/pam-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/standing-privilege/ --- # VPC Service Controls Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. VPC Service Controls is a Google Cloud feature that creates a perimeter around supported services and projects, to reduce the risk of data exfiltration by blocking requests that cross it. Perimeters can run in dry run mode, which logs violations without blocking, or in enforced mode. ## An example, sourced The trusted network condition check asks that sensitive projects sit inside an enforced perimeter. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Workload identity federation: meaning and an example](/glossary/workload-identity-federation/) - [Access Analyzer: meaning and an example](/glossary/access-analyzer/) - [Audit log: meaning and an example](/glossary/audit-log/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/vpc-service-controls/docs/overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/vpc-service-controls/ --- # Workload identity federation Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Workload identity federation lets a workload outside a cloud, such as a CI pipeline or another cloud, obtain short-lived credentials by presenting a token from a trusted identity provider, instead of holding a stored key. Google Cloud and AWS both support it for OIDC identity providers. ## An example, sourced The workload identity check asks that CI systems use it instead of stored secrets. ## Why it matters here On Cloud Security Risk Assessment the term appears in the rubric and on the platform pages; the example above was read on 16 Nov 2026 from the page under Sources. ## Related pages - [Access Analyzer: meaning and an example](/glossary/access-analyzer/) - [Audit log: meaning and an example](/glossary/audit-log/) - [Break-glass account: meaning and an example](/glossary/break-glass-account/) - [All of them](/glossary/) ## Sources - https://cloud.google.com/iam/docs/workload-identity-federation (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/glossary/workload-identity-federation/ --- # What changed in the last 90 days Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Dated changes from AWS, Microsoft, Google Cloud, CIS, NIST and the Cloud Security Alliance that touch the 40 checks, from 18 August to 16 November 2026. Each item names the primary source, says what changed and on what date, and says which check the change relates to, without ranking anyone. - [AWS introduces account access manager for workforce role assignment](/news/aws-account-access-manager/): 5 Nov 2026 - [CIS publishes AWS Foundations Benchmark v7.1.0](/news/cis-benchmarks-november-2026-aws-foundations/): 3 Nov 2026 - [Google Cloud basic roles Admin, Writer and Reader reach GA](/news/gcp-basic-roles-admin-writer-reader-ga/): 29 Oct 2026 - [Defender portal adds posture policies and recommendation exemptions](/news/defender-posture-policies-exemptions/): 28 Oct 2026 - [CSA launches STAR for AI Level 2](/news/csa-star-for-ai-level-2/): 27 Oct 2026 - [NIST drafts guidance on governing ICT risk within enterprise risk](/news/nist-sp-800-221r1-draft/): 22 Oct 2026 - [NIST drafts a CSF 2.0 aligned handbook for executive leadership](/news/nist-sp-800-100r1-draft/): 21 Oct 2026 - [Google Cloud revokes Workforce Identity Federation sessions at GA](/news/gcp-workforce-session-revocation-ga/): 6 Oct 2026 - [IAM Identity Center adds network controls for Identity Store](/news/aws-identity-store-network-access-controls/): 5 Oct 2026 - [Security Command Center reports reasoning engines that can edit IAM](/news/scc-risk-engine-reasoning-engines-iam/): 28 Sep 2026 - [NIST drafts a quick-start guide for AI in CSF analysis](/news/nist-sp-1353-ai-csf-quick-start/): 19 Aug 2026 - [CSA survey links fragmented ownership to manual policy risk](/news/csa-survey-fragmented-operating-models/): 18 Aug 2026 ## Related pages - [The 40 questions of the assessment](/questions/) - [Cloud security assessment tools and options](/options/) - [Guides](/guides/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/ --- # AWS introduces account access manager for workforce role assignment Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS introduced account access manager on 5 November 2026, which it describes as a way to assign existing IAM roles from individual accounts to workforce users and groups in an organization instance of IAM Identity Center. AWS says it gives user-level visibility and works alongside permission sets. The AWS Security Blog says an administrator creates an entitlement to grant a user access to a role in an account, and that the user reaches it from a new portal, the IAM Identity Center portal or with the aws login command. AWS says assignments go to users and groups, which gives accountability for the specific user and role. The feature relates to the workforce sign-in and access evidence checks, since each entitlement ties a person to a role. ## Related pages - [CIS publishes AWS Foundations Benchmark v7.1.0](/news/cis-benchmarks-november-2026-aws-foundations/) - [Google Cloud basic roles Admin, Writer and Reader reach GA](/news/gcp-basic-roles-admin-writer-reader-ga/) - [Defender portal adds posture policies and recommendation exemptions](/news/defender-posture-policies-exemptions/) - [All of them](/news/) ## Sources - https://aws.amazon.com/blogs/security/introducing-aws-account-access-manager-iam-role-flexibility-with-centralized-workforce-access/ (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/aws-account-access-manager/ --- # CIS publishes AWS Foundations Benchmark v7.1.0 Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. CIS announced AWS Foundations Benchmark v7.1.0 in its November 2026 update, posted 3 November 2026. CIS lists expanded AWS Organizations and IAM coverage, enhanced access and credential management guidance, and a new section dedicated to AWS Bedrock security and governance as highlights of the release. The CIS blog post lists the AWS Foundations Benchmark among the Benchmarks updated in the past month and describes the highlights. CIS SecureSuite members can download other formats from CIS WorkBench. Question pages in this assessment cite control numbers by number and short title. Check them against the edition you hold, since a new major edition can renumber controls. ## Related pages - [Google Cloud basic roles Admin, Writer and Reader reach GA](/news/gcp-basic-roles-admin-writer-reader-ga/) - [Defender portal adds posture policies and recommendation exemptions](/news/defender-posture-policies-exemptions/) - [CSA launches STAR for AI Level 2](/news/csa-star-for-ai-level-2/) - [All of them](/news/) ## Sources - https://www.cisecurity.org/insights/blog/cis-benchmarks-november-2026-update (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/cis-benchmarks-november-2026-aws-foundations/ --- # Google Cloud basic roles Admin, Writer and Reader reach GA Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Google's IAM release notes for 29 October 2026 say the Admin, Writer and Reader basic roles are generally available. Google says the legacy Owner, Editor and Viewer roles remain available and maintained, and recommends granting the new roles instead because they support IAM Conditions. The release notes point to Basic roles documentation and say the newer roles support features such as IAM Conditions that the legacy roles do not. For the administrator roles check, a team on Google Cloud should now count both the legacy and the new basic roles when it totals broad assignments. ## Related pages - [Defender portal adds posture policies and recommendation exemptions](/news/defender-posture-policies-exemptions/) - [CSA launches STAR for AI Level 2](/news/csa-star-for-ai-level-2/) - [NIST drafts guidance on governing ICT risk within enterprise risk](/news/nist-sp-800-221r1-draft/) - [All of them](/news/) ## Sources - https://cloud.google.com/iam/docs/release-notes (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/gcp-basic-roles-admin-writer-reader-ga/ --- # Defender portal adds posture policies and recommendation exemptions Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Microsoft's Defender for Cloud release notes for 28 October 2026 say posture policies and recommendation exemptions are in public preview in the Microsoft Defender portal for Azure environments. Microsoft says they allow custom posture evaluations and centralized exemption management with lifecycle tracking. The release notes say the preview lets teams evaluate Azure resources against the Defender for Cloud baseline policy, create their own posture policies, and manage accepted risks at scale. The exemption lifecycle tracking relates to the check on guardrail exceptions, which asks that each has an owner, a reason and an expiry date. ## Related pages - [CSA launches STAR for AI Level 2](/news/csa-star-for-ai-level-2/) - [NIST drafts guidance on governing ICT risk within enterprise risk](/news/nist-sp-800-221r1-draft/) - [NIST drafts a CSF 2.0 aligned handbook for executive leadership](/news/nist-sp-800-100r1-draft/) - [All of them](/news/) ## Sources - https://learn.microsoft.com/en-us/azure/defender-for-cloud/release-notes (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/defender-posture-policies-exemptions/ --- # CSA launches STAR for AI Level 2 Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The Cloud Security Alliance announced STAR for AI Level 2 on 27 October 2026. CSA says it adds third-party-verified AI assurance to the STAR program, built on ISO/IEC 42001 and CSA's AI Controls Matrix, with three certification paths and an attestation path for organizations with SOC 2. CSA's press release names three certification paths that combine ISO/IEC 27001, ISO/IEC 42001 and the AI Controls Matrix, and says a Valid-AI-ted AI-CAIQ submission is required for either certification or attestation. The program certifies governance, not cloud access settings. The AI workload checks in this assessment test the access side that such governance would rely on. ## Related pages - [NIST drafts guidance on governing ICT risk within enterprise risk](/news/nist-sp-800-221r1-draft/) - [NIST drafts a CSF 2.0 aligned handbook for executive leadership](/news/nist-sp-800-100r1-draft/) - [Google Cloud revokes Workforce Identity Federation sessions at GA](/news/gcp-workforce-session-revocation-ga/) - [All of them](/news/) ## Sources - https://cloudsecurityalliance.org/press-releases/2026/10/27/cloud-security-alliance-launches-star-for-ai-level-2-to-strengthen-ai-governance-and-trust (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/csa-star-for-ai-level-2/ --- # NIST drafts guidance on governing ICT risk within enterprise risk Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. NIST released the initial public draft of SP 800-221 Revision 1 on 22 October 2026, titled Enterprise Impact of Information and Communications Technology Risk. The comment period is open through 7 December 2026, and NIST says the document aligns ICT risk management with Cybersecurity Framework 2.0. The NIST publication page lists a publication date of 22 October 2026 and comments due on 7 December 2026, with a comment template. The abstract says the document helps organizations integrate ICT risks into enterprise risk management. A team that feeds assessment results into an enterprise risk register can use the risk register template for that handoff. ## Related pages - [NIST drafts a CSF 2.0 aligned handbook for executive leadership](/news/nist-sp-800-100r1-draft/) - [Google Cloud revokes Workforce Identity Federation sessions at GA](/news/gcp-workforce-session-revocation-ga/) - [IAM Identity Center adds network controls for Identity Store](/news/aws-identity-store-network-access-controls/) - [All of them](/news/) ## Sources - https://csrc.nist.gov/pubs/sp/800/221/r1/ipd (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/nist-sp-800-221r1-draft/ --- # NIST drafts a CSF 2.0 aligned handbook for executive leadership Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. NIST released the initial public draft of Special Publication 800-100 Revision 1 on 21 October 2026. NIST says the draft restructures its Information Security Handbook for executive leadership to follow the organization of the Cybersecurity Framework 2.0 and gives high-level summaries of key cybersecurity topics. The NIST news page lists SP 800-100r1 ipd as released for comments on 21 October 2026, and the publication page names it Information Security Handbook: A Guide for Executive Leadership. The assessment's export is a score per domain and five fixes, a format that suits a summary for leadership. ## Related pages - [Google Cloud revokes Workforce Identity Federation sessions at GA](/news/gcp-workforce-session-revocation-ga/) - [IAM Identity Center adds network controls for Identity Store](/news/aws-identity-store-network-access-controls/) - [Security Command Center reports reasoning engines that can edit IAM](/news/scc-risk-engine-reasoning-engines-iam/) - [All of them](/news/) ## Sources - https://csrc.nist.gov/pubs/sp/800/100/r1/ipd (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/nist-sp-800-100r1-draft/ --- # Google Cloud revokes Workforce Identity Federation sessions at GA Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Google's IAM release notes for 6 October 2026 say revoking active sessions and short-lived credentials for Workforce Identity Federation users across all clients is generally available. Google says administrators can revoke workforce user sessions with the gcloud CLI or the REST API. The IAM release notes list the feature under 6 October 2026 and say it applies to Workforce Identity Federation principals across all clients. It supports the revocation drill check, which asks whether a team can cut off all access of one compromised identity within an hour. A session revocation command removes one step from the written drill. ## Related pages - [IAM Identity Center adds network controls for Identity Store](/news/aws-identity-store-network-access-controls/) - [Security Command Center reports reasoning engines that can edit IAM](/news/scc-risk-engine-reasoning-engines-iam/) - [NIST drafts a quick-start guide for AI in CSF analysis](/news/nist-sp-1353-ai-csf-quick-start/) - [All of them](/news/) ## Sources - https://cloud.google.com/iam/docs/release-notes (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/gcp-workforce-session-revocation-ga/ --- # IAM Identity Center adds network controls for Identity Store Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. AWS announced on 5 October 2026 that IAM Identity Center supports network access controls for Identity Store. AWS says the controls can require Identity Store API requests to arrive through allowed VPC endpoints or source VPCs, and restrict both the Identity Store API and the SCIM API to specific IP ranges. AWS What's New says the controls are optional and turned off by default, that requests AWS services make on your behalf are exempt, and that they are set through the Identity Store API with the AWS SDKs and CLI. The capability is available in all Regions where IAM Identity Center is offered. This bears on the trusted network condition check, which asks whether access to sensitive identity and data stores is limited to approved networks. ## Related pages - [Security Command Center reports reasoning engines that can edit IAM](/news/scc-risk-engine-reasoning-engines-iam/) - [NIST drafts a quick-start guide for AI in CSF analysis](/news/nist-sp-1353-ai-csf-quick-start/) - [CSA survey links fragmented ownership to manual policy risk](/news/csa-survey-fragmented-operating-models/) - [All of them](/news/) ## Sources - https://aws.amazon.com/about-aws/whats-new/2026/10/aws-identity-store-network-controls/ (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/aws-identity-store-network-access-controls/ --- # Security Command Center reports reasoning engines that can edit IAM Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Google's Security Command Center release notes for 28 September 2026 say Risk Engine now detects and reports reasoning engines that can modify IAM policies and perform lateral movement. Google says these findings are generated as a toxic combination, and Security Command Center lists the feature as a new capability. The Security Command Center release notes list the feature under 28 September 2026. They describe the findings as generated as a toxic combination. This maps to the AI workload checks on tool permissions and on each agent having its own identity, which ask whether an agent can change IAM at all. ## Related pages - [NIST drafts a quick-start guide for AI in CSF analysis](/news/nist-sp-1353-ai-csf-quick-start/) - [CSA survey links fragmented ownership to manual policy risk](/news/csa-survey-fragmented-operating-models/) - [AWS introduces account access manager for workforce role assignment](/news/aws-account-access-manager/) - [All of them](/news/) ## Sources - https://cloud.google.com/security-command-center/docs/release-notes (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/scc-risk-engine-reasoning-engines-iam/ --- # NIST drafts a quick-start guide for AI in CSF analysis Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. NIST published the initial public draft of Special Publication 1353 on 19 August 2026, a quick-start guide for using artificial intelligence for Cybersecurity Framework 2.0 analysis and reporting. The comment period ran to 15 October 2026 and is now closed, according to NIST's publication page read on 16 Nov 2026. The NIST publication page for SP 1353 lists a publication date of 19 August 2026 and comments due on 15 October 2026, and marks the public comment period as closed. The draft concerns analysis and reporting with CSF 2.0, not cloud configuration. It matters to the assessment only where a team uses AI to summarize findings, in which case the AI workload checks on identity and logging apply to that tooling too. ## Related pages - [CSA survey links fragmented ownership to manual policy risk](/news/csa-survey-fragmented-operating-models/) - [AWS introduces account access manager for workforce role assignment](/news/aws-account-access-manager/) - [CIS publishes AWS Foundations Benchmark v7.1.0](/news/cis-benchmarks-november-2026-aws-foundations/) - [All of them](/news/) ## Sources - https://csrc.nist.gov/pubs/sp/1353/ipd (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/nist-sp-1353-ai-csf-quick-start/ --- # CSA survey links fragmented ownership to manual policy risk Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The Cloud Security Alliance published a survey on 18 August 2026 saying fragmented operating models across teams, tools and environments, still reliant on manual processes, have made manual policy management a production risk. CSA's summary names limited visibility and fragmented ownership as the causes. The CSA press release list, read 16 Nov 2026, shows the survey under the headline that legacy operating models cannot keep pace with IT complexity. The excerpt says fragmented ownership and limited visibility have made manual policy management a production risk. For the assessment, the finding supports the guardrail checks that ask for policies as code and named owners for exceptions. ## Related pages - [AWS introduces account access manager for workforce role assignment](/news/aws-account-access-manager/) - [CIS publishes AWS Foundations Benchmark v7.1.0](/news/cis-benchmarks-november-2026-aws-foundations/) - [Google Cloud basic roles Admin, Writer and Reader reach GA](/news/gcp-basic-roles-admin-writer-reader-ga/) - [All of them](/news/) ## Sources - https://cloudsecurityalliance.org/press-releases (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/news/csa-survey-fragmented-operating-models/ --- # 30 questions, answered from the data Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Answers to common questions about the 40-question cloud access assessment: how long it takes, what it covers, how the score and the five fixes are calculated, where answers go and how it differs from an audit. Each answer is short, names its source and links to the page behind it. **How long does the cloud security risk assessment take?** About 20 minutes to answer the 40 questions for a team that knows its estate. Verifying each answer against a console takes longer, and depends on the number of accounts. Each [question page](/questions/root-account-mfa/) gives the AWS, Azure and Google Cloud test. **What does the assessment cover?** It covers the cloud access layer in six domains: identities and entitlements, network reachability, data perimeter, AI workloads, guardrails and drift, and evidence. Workload vulnerabilities and posture scanning are outside it. See the [checklist](/cloud-security-assessment-checklist/) for all 40 checks. **Does the assessment rank anything?** No. It scores your own answers and ranks no provider, tool or vendor. The [options](/options/) page lists assessment tools in alphabetical order within each kind, with each maker's own description and the date it was read. **Where do my answers go?** Nowhere. Answers stay in the browser, the page sets no cookies and runs no analytics, and the PDF and XLSX exports are made in the browser. Closing the tab ends the session unless you saved an export. **How is the domain score calculated?** The sum of answer value times question weight, divided by the sum of weights, times 100. Values are 1 for yes, 0.5 for partly and 0 for no or not known. The [methodology](/methodology/) and the [cloud risk assessment](/cloud-risk-assessment/) guide show a worked example. **How are the five fixes chosen?** The tool takes the questions with the highest weight times one minus the answer value. Ties go to lower effort first, then to question order. The result is a short list one team can finish, with each fix written as one imperative sentence. **Why are there 40 questions?** Forty keeps the run near 20 minutes while giving each of six domains six to eight checks. Domain sizes are 8, 7, 7, 6, 6 and 6. The count is a design choice stated on the [methodology](/methodology/) page, not a standard. **What do the weights mean?** A weight from 1 to 3 records how much a wrong answer moves risk. Twelve of the 40 questions weigh 3, and the weights sum to 84. Domain weights are separate: 25, 20, 20, 15, 10 and 10, which sum to 100. **Why does not known score as no?** A control nobody can show offers the same exposure as a missing one. Scoring unknown as no keeps the result honest and puts the missing evidence on the fix list, where an owner can close it. **Does it work for AWS, Azure and Google Cloud?** Yes. Every question applies to all three, and each question page gives one test per cloud. The [AWS](/aws-security-checklist/), [Azure](/azure-security-checklist/) and [Google Cloud](/google-cloud-security-checklist/) checklists pick the checks with the most direct tests. **Which controls do the questions come from?** From CIS Benchmarks, the Microsoft cloud security benchmark, NIST SP 800-53 Rev. 5 and provider guidance from AWS and Google Cloud. CIS and CSA texts are referenced by number and short title only, because they are licensed. NIST may be quoted. **Why include AI workloads?** AI agents and pipelines act with cloud credentials, so what they can call and read is an access question. The domain has six checks and a weight of 15, covering identity, permissions, approved models, data reach, human approval and logging. **Is this a vulnerability scan?** No. It does not scan workloads for vulnerabilities or posture settings. Those jobs use other tools, and general security platform rankings are not covered here. The assessment asks about access: who can sign in, what they can reach and what is logged. **How is an assessment different from an audit?** An assessment estimates risk and ranks gaps. An audit tests whether a control operated and keeps dated evidence. The [cloud security audit](/cloud-security-audit/) guide and the [audit checklist](/cloud-security-audit-checklist/) cover the second. **How often should we run it?** After each batch of fixes and at least every quarter. Two runs show a trend, and each export is a dated record. Keep the same scope each time so scores compare. **Who should answer the questions?** One person each from identity, network and data or platform, so no domain is answered by guesswork. A single security lead can start the run, but the owners should confirm answers in their own area. **What is a good score?** This site sets no pass mark. Read each domain score against the [maturity model](/cloud-security-maturity-model/) bands and decide a target per domain. A high score in evidence does not offset a low one in identities. **What is in the risk register?** The XLSX has 15 columns, from risk ID and check ID to gap, owner, due date and evidence link. The assessment fills 8 and calculates the gap; you add the rest. See the [register template](/risk-register-template/). **Can I cite the assessment?** Yes. The cite line on the method page reads: Cloud Security Risk Assessment, the 40-question assessment, edition 16 Nov 2026, with the methodology address. Cite the edition date, since questions change between editions. **What does phishing-resistant MFA mean here?** Authentication where the credential is bound to the real site or device, such as passkeys, FIDO2 security keys or certificates. Microsoft lists these under its authentication methods (read 16 Nov 2026). Text message and app codes can be phished. **Why check the metadata service?** On AWS, a request-forgery bug can read credentials from the instance metadata address. Requiring IMDSv2 blocks many of those. Azure and Google Cloud have equivalent concerns, which the [question page](/questions/metadata-service-hardened/) covers. **What is a break-glass account?** An emergency administrator account for when normal sign-in fails. Microsoft calls it an emergency access account. It needs long unique credentials, alerting on every use and a test at least twice a year. **Are free provider tools enough?** They answer part of the questions. AWS Trusted Advisor, Defender for Cloud and Security Command Center each report on a subset of checks, and each has plan conditions listed on the [options](/options/) page. None covers every domain. **Does Act offer an assessment?** Yes. Act's page, read 16 Nov 2026, describes a free access attack simulation that is read-only and deployed in under 5 minutes. It is listed on the [options](/options/) page beside the other assessments. **What is the CIS Benchmark version?** CIS announced AWS Foundations Benchmark v7.1.0 in its November 2026 update (read 16 Nov 2026). Question pages cite control numbers from the edition named on each page, so confirm against the edition you hold. **Can I use it for a small team?** Yes. A team with one account can answer all 40 and verify most in an afternoon. Start with the [15-item list](/cloud-security-checklist/) if the account is new. **What counts as partly?** The control is in place in some accounts or for some identities, but not all. Do not use it for plans or pilots, which count as no. Record which accounts are covered in the register. **How do I fix the lowest domain first?** Sort the fixes by gap and start with the low-effort ones that weigh 3, such as root account MFA, open admin ports, public storage and audit logging. Then move to the high-effort items on the list. **Where are corrections reported?** Write to editor@cloudsecurityriskassessment.com with the page, the sentence and the source. Confirmed errors are fixed on the page and logged with the date. **How current are the facts?** Provider facts carry the date they were read, 16 Nov 2026 for this edition. The news page lists dated changes from AWS, Microsoft, Google, CIS, NIST and CSA that affect the checks. ## Related pages - [The 40-question assessment: the data, downloadable](/datasets/) - [About](/about/) - [Editorial policy](/editorial-policy/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/faq/ --- # The 40-question assessment: how the cloud access score is calculated Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The assessment scores a team's answers to 40 questions in six domains. Each answer is worth 1, 0.5 or 0, each question weighs 1 to 3, and a domain score is the weighted share of yes answers from 0 to 100. The five fixes are the largest weighted gaps, with ties broken by lower effort. Every file below is this site's own data, edition 16 Nov 2026, under CC BY 4.0. The [methodology](/methodology/) explains how it was made. ## Downloads - [Cloud access assessment questions (CSV)](/datasets/cloud-access-assessment-questions-2026-11-16.csv): 40 rows, edition 16 Nov 2026 - [Cloud access assessment questions (JSON)](/datasets/cloud-access-assessment-questions-2026-11-16.json): The 40 questions with domain, weight, answer values, fix, effort and the control each comes from. - [Cloud security assessment options (CSV)](/datasets/cloud-security-assessment-options-2026-11-16.csv): 15 rows, edition 16 Nov 2026 - [Cloud security assessment options (JSON)](/datasets/cloud-security-assessment-options-2026-11-16.json): Provider tools, open source scanners and vendor-run assessments, the same fields for each, with source and read date. ## Cite this **Cite this.** Cite this line: 'Cloud Security Risk Assessment, the 40-question assessment, edition 16 Nov 2026, https://cloudsecurityriskassessment.com/methodology/' ## Changelog - 16 Nov 2026: first edition. Every row read from the source linked in the row on 16 Nov 2026. ## Related pages - [About](/about/) - [Editorial policy](/editorial-policy/) - [Corrections](/corrections/) ## Sources - CIS AWS Foundations Benchmark v5.0: https://www.cisecurity.org/benchmark/amazon_web_services (read 16 Nov 2026) - Microsoft cloud security benchmark: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - Google Cloud security best practices: https://cloud.google.com/iam/docs/best-practices-for-managing-service-account-keys (read 16 Nov 2026) - Microsoft cloud security benchmark: https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-identity-management (read 16 Nov 2026) - AWS IAM Access Analyzer: https://docs.aws.amazon.com/IAM/latest/UserGuide/what-is-access-analyzer.html (read 16 Nov 2026) - AWS Security Hub CSPM: https://docs.aws.amazon.com/securityhub/latest/userguide/what-is-securityhub.html (read 16 Nov 2026) - AWS Trusted Advisor security checks: https://docs.aws.amazon.com/awssupport/latest/user/trusted-advisor.html (read 16 Nov 2026) - Google Cloud IAM recommender: https://cloud.google.com/iam/docs/recommender-overview (read 16 Nov 2026) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/datasets/ --- # About Cloud Security Risk Assessment: a scored review of the cloud access layer Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cloud Security Risk Assessment is a 40-question tool that scores the access layer of an AWS, Azure or Google Cloud estate. It publishes the questions, the weights and the arithmetic, a checklist and guide for each topic, a directory of assessment options, and dated news. It ranks no provider, tool or vendor. The site has one job: let a team assess who and what can reach its cloud accounts and data, in about 20 minutes, and leave with a scored report. The questions are written once and apply to AWS, Azure and Google Cloud, with a test per cloud on each question page. Workload vulnerabilities, posture scanning and security platform rankings are outside the scope and get one sentence at most. The directory of options lists provider tools, open source scanners and vendor-run assessments side by side, in alphabetical order within each kind and with each maker's own description. ## Related pages - [Editorial policy](/editorial-policy/) - [Corrections](/corrections/) - [Contact](/contact/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/about/ --- # How Cloud Security Risk Assessment is written Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cloud Security Risk Assessment publishes the 40 questions, their weights and the scoring arithmetic in full. Each page answers its question in a short summary under the headline, names its numbers and lists its sources with the date read, so a reader can check any claim. Provider statements are labelled as the provider's own. Licensed texts from CIS and CSA are referenced by number and short title, never quoted. NIST text may be quoted. A capability that was not found is written as none found on a date, never as an absence. Pages are reviewed when a source changes, and the review date is shown on each page. ## Related pages - [Corrections](/corrections/) - [Contact](/contact/) - [Privacy](/privacy/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/editorial-policy/ --- # Corrections and the correction rule Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Corrections go to editor@cloudsecurityriskassessment.com. Include the page address, the sentence you think is wrong and the document that shows the right figure. A confirmed error is fixed on the page, the review date changes, and a dated note describing the change is added to the corrections log. Corrections to a question, weight or answer value create a new edition date, because scores from different editions are not comparable. ## Related pages - [Contact](/contact/) - [Privacy](/privacy/) - [Terms](/terms/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/corrections/ --- # Contact Cloud Security Risk Assessment Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Write to editor@cloudsecurityriskassessment.com with corrections, questions about a page or requests to cite or reuse the assessment. Please include the page address and, for a correction, the sentence and the source that shows the right figure. Messages are read by the editor. The site does not take calls. ## Related pages - [Privacy](/privacy/) - [Terms](/terms/) - [How the assessment is scored](/methodology/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/contact/ --- # Privacy Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. Cloud Security Risk Assessment sets no cookies and runs no analytics script. Your answers stay in the browser and are not sent to a server, and the PDF and XLSX exports are made in the browser on your device. The web server keeps ordinary access logs, and nothing else is collected about visitors. Closing the page clears the answers unless you saved an export. Treat the saved files as you would any record of your security gaps. ## Related pages - [Terms](/terms/) - [How the assessment is scored](/methodology/) - [Cloud access risk register template (XLSX)](/risk-register/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/privacy/ --- # Terms of use Facts read 16 Nov 2026. Page reviewed 16 Nov 2026. The content of Cloud Security Risk Assessment is general information about cloud access control and is provided as is. It is not legal, audit or professional advice, and the pages describe provider documentation as read on the date shown. Check current provider documentation and your own evidence before you act on a score. A score reflects the answers given. The site cannot verify them. The questions may be cited with the edition date. ## Related pages - [How the assessment is scored](/methodology/) - [Cloud access risk register template (XLSX)](/risk-register/) - [30 questions about cloud access risk assessment](/faq/) Last reviewed 16 Nov 2026. Canonical: https://cloudsecurityriskassessment.com/terms/