What is AI governance?
AI governance is the set of policies, roles and controls an organization uses to decide which AI systems it runs, who is responsible for each one, what each one may do, and how all of that is evidenced. For AI agents it adds ownership, permitted actions and a record of actions to the usual concerns about models.
Also called AI agent governance, AI governance platform. Last reviewed .
Why it matters
Three documents set most of the requirements a governance lead works from. The EU AI Act is law in the European Union and places obligations on providers and deployers of AI systems according to risk. The NIST AI Risk Management Framework organizes the work into four functions: Govern, Map, Measure and Manage. ISO/IEC 42001 is the certifiable management system standard for AI.
All three were written with models and applications in mind. Agents raise questions those texts answer only in general terms: which person answers for an agent that acts on its own, what it was permitted to do, and what record shows what it did.
Related terms
- AI agent inventory
- An AI agent inventory is the list governance applies to.
- Human of Record
- A Human of Record is how governance assigns responsibility for one agent to one person.
- Decision record
- A decision record is the evidence governance produces and auditors read.
- AI security posture management (AI-SPM)
- AI security posture management supplies the technical findings that governance decisions rest on.
How Guard0 handles it
Guard0 gives every agent a named Human of Record, a policy and a decision on the record. Findings map to the EU AI Act, ISO 42001, the NIST AI RMF, the OWASP Top 10 lists and many more, and a policy can post to Slack, open a Jira ticket or page the on-call person when an agent breaks a rule.
Read the guide: EU AI Act Compliance for AI Agents: Article 14 and Beyond