Use a task-to-permission matrix to give an AI agent bounded access, separate drafts from sends, and test denied actions before release.