Apple changes full-disk access permissions to curb abuse from AI agents

by | Oct 22, 2026 | Technology

Apple changes full-disk access permissions to curb abuse from AI agents

Apple announced changes to its macOS privacy settings targeting the full-disk access permission, which grants applications broad access to files stored on a user’s computer. The move follows a high-profile incident in which Meta’s AI assistant Muse sent a notification to a technology columnist referencing messages he had not knowingly authorized the application to access.

The controversy began when columnist Jason Aten reported that Muse had referenced a private message thread without his explicit permission. Aten stated he had not granted Muse access to his messages and assumed they were protected. The incident generated significant social media discussion about the risks of granting AI systems broad permissions to access calendars, emails, messages, and other personal data.

Meta’s Chief Technology Officer David Singleton initially disputed claims that Muse improperly accessed messages, stating that the application can only read Messages content if a user manually enables two separate permissions: the macOS system-level full-disk access setting and a specific Messages connector within Muse. Security researcher Patrick Wardle challenged this assertion, explaining that full-disk access inherently allows applications to read all non-root files on a system, including messages and browsing history, regardless of additional settings.

Apple’s statement emphasized that some developers are using full-disk access in ways that could expose sensitive user information without adequate user awareness. The company noted that as AI systems become more autonomous and capable, the risks associated with broad system-level access will increase substantially. Apple stated it is committed to ensuring users understand these risks before granting such permissions.

The timing of Apple’s announcement follows other security concerns involving Muse, including a disclosed vulnerability that could allow attackers to take control of the AI assistant, and Amazon’s decision to block the application from its platform. Security experts and observers suggest that the extraordinary access level required for Muse to function as designed may outweigh its benefits, and users are advised to configure permissions with particular care.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI