Apple Tightens macOS Full Disk Access Rules Amid AI Agent Privacy Concerns

Apple has announced it will implement additional safeguards around the “Full Disk Access” permission on macOS, a move driven by escalating privacy risks posed by increasingly autonomous AI agents. The decision follows a recent incident where Meta’s Muse app was accused of accessing private user messages without explicit consent, a claim the company has disputed.

The Full Disk Access feature was originally designed to facilitate system backups, but Apple now warns that AI agents can leverage this capability to read files, mail, messages, and browsing histories with minimal oversight. Jason Aten, an Inc. columnist, reported that Muse appeared to know the contents of his private messages despite his belief that he had not granted permission. His report sparked broader concerns about the trust users place in desktop-based AI tools that operate with deep system integration.

Apple emphasized in a developer-focused blog post that some applications are utilizing this access in ways that expose sensitive user data without clear understanding or consent. “As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially,” the company stated, adding that future updates will require “very explicit user action” for anyone wishing to grant an app such extraordinary permissions.

This policy shift also aligns with a recent Wired report highlighting a vulnerability in ChatGPT’s Mac application that could have allowed hackers to extract sensitive information. Apple did not respond to requests for comment regarding the specifics of the upcoming feature changes.