Every AIUC-1 requirement ID with a one-line topic in our own words and how AutoGovern's anchor controls (NIST AI RMF, ISO/IEC 42001 Annex A, EU AI Act) cover it. Read the official standard for the requirement text.
55 IDs: 8 full mapping · 35 partial · 7 not covered by our anchors · 5 retired · 4 disputed between our analysts.
AIUC-1 Q4-2026 release of 2026-10-15; next scheduled release 2027-01-15. AIUC-1 publishes a release each quarter. Retired requirements stay listed with their IDs. Researched on 2026-10-16.
Signed in? See which of these your registered controls already meet under Account → Workspace → Standards readiness.
Control and requirement IDs belong to the Cloud Security Alliance (AICM) and the Artificial Intelligence Underwriting Company (AIUC-1). Descriptions and mappings are AutoGovern's own; read the official standard for the requirement text.
Official pages: CSA AI Controls Matrix (AICM) v1.1 · AIUC-1 · the AIUC-1 standard
Two analysts mapped every item independently to the anchor controls; on disagreement the lower strength and the union of anchor controls were kept, and the item is marked disputed. This is AutoGovern's analysis, not an official CSA or AIUC crosswalk. "Not covered" means our AI-governance anchor controls do not address the item (most cloud-infrastructure controls). It does not mean an organization fails it.
Information duties to users exist; a customer-facing data use, retention and training policy is narrower.
Customer expectations and information duties overlap; output ownership and deletion choices are not addressed.
Use is bounded to intended scope; context-based limits on agent data access are not specified.
Confidentiality and extraction attacks are in scope; preventing leakage of company secrets is narrower.
Cross-customer data isolation is a multi-tenancy control not in our anchors.
Privacy risk is examined; technical prevention of personal data in outputs and logs is not required.
Third-party IP risk is mapped; preventing outputs that infringe rights is a different, output-side duty.
Detecting exposed secrets and credentials is not addressed by anchors.
Independent assessment and privacy evaluation exist; quarterly expert third-party testing is more demanding.
Independent assessors and security evaluation overlap; quarterly external adversarial testing is not required.
Detecting and responding to attacks is required; prompt-injection monitoring specifically is not.
Limiting public technical disclosures to reduce targeting is not covered.
Model extraction is an evaluated threat; rate limiting and anti-scraping defences are not specified.
Adversarial input resilience overlaps; real-time automated moderation is a specific mechanism not required.
Intended-use limits and override capability overlap; technical enforcement of agent permissions is not required.
Role-based user and admin access control is not addressed by anchors.
Hosting environment encryption and access hardening fall outside the anchors beyond a general cybersecurity duty.
Confidentiality attacks are in scope; output obfuscation and fidelity limits are not specified.
Secure patterns in generated code are not addressed by anchors.
Context-based identification and severity characterisation of AI risks is required.
Documented testing against criteria before deployment is required.
Safety and bias are evaluated and treated; runtime safeguards against toxic or deceptive outputs are more specific.
Scope is specified and documented; technical blocking of off-topic responses is not required.
Risk mitigation is required; use-case-specific output safeguards are more specific.
Security of the system is required; preventing exploitable content in outputs is narrower.
Human oversight is designed in; an alerting route for high-risk outputs is a specific mechanism.
Ongoing production and post-market monitoring of risks is required.
Feedback channels and intervention exist; real-time collection and actioning is not required.
Independent assessment and safety evaluation overlap; quarterly expert external testing is more demanding.
Accuracy is measured and limits documented; safeguards against fabricated content are more specific.
Intended-use limits and override overlap; technical controls on tool invocation are not required.
Independent assessment and reliability evaluation overlap; quarterly expert external testing is more demanding.
Incident response and communication exist; breach-specific plan with external support partners is narrower.
Incident response covers harm; a named-owner plan for customer harm with outside support is more specific.
Incident response exists; a plan specific to financial loss from fabricated outputs is narrower.
Roles and deployment approval exist; listing which changes need sign-off with evidence is not required.
Data management considers security; documented storage security practices across environments are more specific.
Third-party and upstream model risk assessment and supplier alignment are required.
Periodic review of risk process and policy exists; reviews of all key processes with approvals is broader.
Logging third-party API connections and data access is not covered by anchors.
AI policy and responsible-use processes overlap; explicit acceptable-use rules are not required.
Compute resource locations are documented; all data processing locations across vendors are not explicit.
Applicable AI legal requirements and a compliance strategy must be documented.
A quality management system for AI is directly required.
Automatic event logs for traceability with retention are required.
Disclosure that users are interacting with AI is directly required.
Technical and user documentation exist; a transparency policy with a model-card repository is not required.
Independent assessment and accountability risk review overlap; quarterly external testing is more demanding.
Foreseeable misuse must be assessed and mitigated; cyber-misuse guardrails are not specifically required.
Foreseeable misuse must be assessed and mitigated; chemical, biological, radiological, nuclear guardrails are not specific.
Control and requirement IDs belong to the Cloud Security Alliance (AICM) and the Artificial Intelligence Underwriting Company (AIUC-1). Descriptions and mappings are AutoGovern's own; read the official standard for the requirement text.
Official pages: CSA AI Controls Matrix (AICM) v1.1 · AIUC-1 · the AIUC-1 standard
Two analysts mapped every item independently to the anchor controls; on disagreement the lower strength and the union of anchor controls were kept, and the item is marked disputed. This is AutoGovern's analysis, not an official CSA or AIUC crosswalk. "Not covered" means our AI-governance anchor controls do not address the item (most cloud-infrastructure controls). It does not mean an organization fails it.