Summary
Nuggets launched the Authority Control Plane (ACP), an enterprise platform that defines, delegates, enforces, and proves what autonomous AI agents are authorized to do. Announced in late August 2026, the ACP sits in the execution path between agents and enterprise systems, evaluating verified identity, delegated authority, policy, intent, and runtime context before each action, with real-time revocation and an estate-wide kill switch that fails closed.
What changed
Nuggets introduced the Authority Control Plane: an in-path authorization layer that checks an agent's identity, delegated authority, organizational policy, and runtime context before an action executes, lets admins scope reachable systems, spending thresholds, approval triggers, and expiry, and supports real-time issue/delegate/revoke plus an estate-wide kill switch that fails closed.
Why it matters
Existing IAM can say who an agent is and log what it did, but not confirm whether a specific action was authorized at the moment it happened. As agents begin initiating transactions and modifying systems, a per-action authorization and revocation layer becomes essential enterprise infrastructure, sharpened by EU AI Act high-risk obligations taking effect in August 2026.
Evidence excerpt
The Authority Control Plane sits in the execution path between AI agents and the systems they interact with; when an agent attempts an action, the ACP evaluates its identity, authority, organizational policy, intent, and runtime context before the action proceeds, with authority issued, delegated and revoked in real time and an estate-wide kill switch failing closed.