AI Governance

AI governance establishes how an organization directs and oversees its use of AI. It connects permitted purposes, decision authority, required evidence, and follow-through across the system's life. The scope includes the application and people using it, not only the model.

A score does not grant permission

In a fictional shop, an assistant passes tests for drafting refund explanations. That result does not authorize it to transfer money. Governance identifies who can approve that expanded use and what additional evidence is needed.

A service owner may allow staff-only drafts while withholding payment permissions. The decision should identify its scope and conditions so operators can apply it.

Keep the decision connected to operations

A changed policy source or new action can invalidate earlier evidence. Record deployed uses, monitor relevant failures, and define when review is required. A governance document has little practical value if nobody can restrict an unsafe use or resolve an outstanding condition.

Reference: NIST AI RMF Core.