Control catalog
Controls organized by NIST AI RMF function. Each control declares which risk tier(s) must implement it and the ISO 42001 clause(s) it satisfies. The control bundle a use case must implement is derived from its tier.
GOVERN · 4 controls
A culture of risk management is cultivated and present.
- AI-GV-1.1AI policy approved by senior leadershipTier 1Tier 2Tier 3
A written AI use policy approved at the C-suite or board level, reviewed annually.
GOVERN 1.1 · ISO 42001: 5.2, 5.3
- AI-GV-1.2AI risk owner assigned per use caseTier 1Tier 2Tier 3
Each registered use case has a named owner accountable for its risk posture.
GOVERN 1.2 · ISO 42001: 5.3
- AI-GV-2.1AI training delivered to operators and stewardsTier 1Tier 2
Personnel with AI responsibilities have completed role-appropriate training in the last 12 months.
GOVERN 2.1 · ISO 42001: 7.2, 7.3
- AI-GV-3.1Pre-deployment review board approves Tier 1 systemsTier 1
Tier 1 systems require AI Review Board sign-off before going live; minutes retained.
GOVERN 3.1 · ISO 42001: 8.2, 8.3
MAP · 3 controls
Context is recognized and risks related to context are identified.
- AI-MP-1.1Use-case registered with intended purpose and stakeholdersTier 1Tier 2Tier 3
Registry entry includes purpose, business owner, affected stakeholder groups, and data inputs.
MAP 1.1 · ISO 42001: 6.1.2, 8.2
- AI-MP-2.1Impact assessment documentedTier 1Tier 2
Documented assessment of impact on individuals, groups, and the organization, including foreseeable misuse.
MAP 2.1 · ISO 42001: 6.1.4
- AI-MP-3.1Data lineage and provenance recordedTier 1Tier 2
Source datasets, licensing, collection method, and known biases of training/grounding data are documented.
MAP 3.1 · ISO 42001: 7.5, 8.1
MEASURE · 4 controls
Identified risks are assessed, analyzed, or tracked.
- AI-MS-1.1Validity and reliability metrics defined and trackedTier 1Tier 2
Documented accuracy / quality metrics with thresholds, monitored on a published cadence.
MEASURE 1.1 · ISO 42001: 9.1
- AI-MS-2.1Bias and fairness testing across protected groupsTier 1
Disparate-impact testing across protected attributes performed prior to deployment and on a recurring basis.
MEASURE 2.1 · ISO 42001: 6.1.4, 9.1
- AI-MS-3.1Adversarial robustness testingTier 1
Red-team or adversarial evaluation against documented threat model; findings tracked.
MEASURE 3.1 · ISO 42001: 8.2
- AI-MS-4.1Privacy and data-protection review completedTier 1Tier 2
DPIA or equivalent privacy review covering training data and prompts; signed off by privacy lead.
MEASURE 4.1 · ISO 42001: 6.1.4, 8.2
MANAGE · 4 controls
Risks are prioritized and acted upon based on a projected impact.
- AI-MG-1.1Human override and contestability pathTier 1
Affected individuals can contest an AI-driven decision; documented escalation path.
MANAGE 1.1 · ISO 42001: 8.2, 10.2
- AI-MG-2.1Production monitoring and drift detectionTier 1Tier 2
Operational telemetry on input distribution, output rates, and quality metrics with alert thresholds.
MANAGE 2.1 · ISO 42001: 9.1
- AI-MG-3.1Incident response runbook for AI failuresTier 1Tier 2
Documented procedure for hallucination, bias, leak, or downtime incidents; rehearsed annually.
MANAGE 3.1 · ISO 42001: 10.2
- AI-MG-4.1Decommissioning and rollback planTier 1
Plan to disable, replace, or roll back the system; data retention/deletion specified.
MANAGE 4.1 · ISO 42001: 8.2