Protection | AI Agent Security

Protection is the difference between watching AI agent activity and governing it. Rules can be scoped tightly enough to permit AI coding broadly while still blocking a specific agent, a specific file path, or a specific environment, so you are not forced to choose between productivity and control. And because policy can be simulated first, you can quantify the impact before your users feel it.

The Protection section has two pages:

• Policies: the policies that audit or block AI agent actions, including the curated BeyondTrust Managed set, and the builder you use to write your own.
• Exception requests: the queue of blocked actions someone has asked to unblock, where you approve or deny each one for a limited time.