What this check asks
Organization's roleselect · required
System categoryselect · required
A proportionate AI risk-management framework, policies, procedures, roles, and training are documentedboolean
Due diligence and foreseeable adverse-impact assessment were completed before deploymentboolean
Use restrictions, safeguards, access controls, and escalation routes match the assessed risksboolean
Users are clearly told they are interacting with AI and receive useful capability and limitation informationboolean
Human oversight, intervention, review, and accountability are defined for the useboolean
Harmful uses and impacts are monitored and feedback, incidents, and corrective action are managedboolean
Fairness, equity, human rights, accessibility, and affected groups are consideredboolean
The system is tested for intended tasks, contexts, failure modes, limitations, and robustnessboolean
Cybersecurity, privacy, confidential-information, data, and supplier controls are documentedboolean
The manager receives and shares information needed to avoid governance gaps with developers and suppliersboolean
Governance review cycleselect