Apple Tightens macOS Full Disk Access, Citing AI Agent Risks
TL;DR
- Apple has not specified when the new controls ship, leaving third-party AI developers in compliance limbo as macOS agentic deployments accelerate.
- Meta Muse was the proximate trigger: users reported it read private messages without explicit consent, compounding Meta's July Muse Image pullback over unauthorized Instagram account linking.
- Apple's direct language states 'as AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially,' signaling an ongoing governance posture rather than a one-time fix.
Apple is adding new controls around macOS Full Disk Access, and is naming AI agents as the reason.
In a statement to TechCrunch, the company said: 'Some developers are using Full Disk Access in ways that could put users at risk, exposing everything on their systems…without users' full knowledge and understanding.' Users who 'genuinely wish to grant an app this extraordinary level of access,' Apple said, will be able to do so only with 'very explicit user action.'
Full Disk Access is a macOS permission that covers files, mail, messages, and browsing history. It was originally built so backup software could work. Reporter Sarah Perez pointed to two recent incidents behind Apple's framing: Inc. columnist Jason Aten said Meta's Muse app read his private messages without permission, which Meta disputed, and Wired documented a flaw in the ChatGPT Mac app that could have let attackers reach sensitive data.
Apple's own words carry the judgment: 'As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially. We are committed to ensuring users clearly understand these risks before granting such access, so they can make informed decisions about their own data and privacy.'
The statement names no macOS version, sets no date, and quotes no Apple executive. It lands the same day Microsoft's annual threat report argued AI has tipped near-term advantage to attackers, and continues a run of tracked Apple coverage on our radar this week.
What others are reporting
-
9to5Mac Read →
Names Meta Muse and OpenAI's Dots as the always-on agents that triggered Apple's move, and links to documented 'horror stories' grounding the policy in concrete user incidents.
Some developers are using Full Disk Access in ways that could put users at risk, exposing everything on their systems
-
MacRumors Read →
Highlights that Apple gave no implementation timeline, and carries Apple's own forward-looking framing about autonomous agent risks growing over time.
As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially.
-
AppleInsider Read →
Cites an OpenAI Mac app bug that could have allowed malicious command execution alongside Muse, widening the named compliance targets beyond Meta alone.
Some developers are using Full Disk Access in ways that could put users at risk, exposing everything on their systems.
-
The National News Read →
Adds that Meta scrapped Muse Image in July over unauthorized Instagram account linking, establishing a pattern of Meta AI privacy failures, and notes Mac revenue surged 29% YoY giving Apple market incentive to act.
Some developers are using full disk access in ways that could put users at risk, exposing everything on their systems.
Shared on Bluesky by 1 AI expert
Originally reported by techcrunch.com
Read the original article →Original headline: Apple Tightens macOS Full Disk Access Controls, Citing Growing Risks From Autonomous AI Agents