Apple Enhances macOS Security to Mitigate Privacy Risks Associated with AI Agents
Apple's Warnings Regarding Full Disk Access Permissions
Apple has warned against the risks of granting AI agents Full Disk Access permissions in macOS. The company explained that these elevated privileges allow AI tools to access personal files, emails, messages, and browsing history, potentially exposing users' and their contacts' data to severe security risks without their full awareness.
AI Agent Behavior and User Concerns
Applications like OpenClaw, Dots, and Muse request Full Disk Access to execute complex, interactive tasks. Despite the power of these tools, privacy concerns have prompted some users to run AI agents on secondary, isolated devices rather than their primary computers. These anxieties intensified following an incident reported by tech writer Jason Aten, where the Muse app accessed his messages—a situation Meta justified as a consequence of the user opting into message synchronization.
Apple's Plan to Tighten Controls in macOS
Apple plans to introduce clearer steps and stricter procedures that require explicit user confirmation before granting any app Full Disk Access. The company emphasized that these updates are vital as AI agents become more autonomous, aiming to help users understand the extent of permissions they grant and make informed decisions about their privacy, though no release date for these controls has been announced yet.