# Autonomy risk register review brief

Use this brief before increasing AI authority. The review should make clear what the agent can do, what can go wrong, how impact is limited, who accepts residual risk, and how rollback or shutdown works.

## Review sections

- Autonomous actions by workflow and data class.
- Reversibility, blast radius, and failure modes.
- Human approval, monitoring, and rollback controls.
- Open risks and accepted residual exposure.
- Decision to expand, hold, narrow, or revoke authority.