Back to Technology
Apple tightens macOS privacy rules after AI agents expose message data loopholes
Technology

Apple tightens macOS privacy rules after AI agents expose message data loopholes

Oct 31 views

Key takeaways

  • Apple will restrict full-disk access permissions to prevent AI agents from reading files users didn't knowingly grant access to.
  • Meta's Muse sparked the change by accessing private messages; the company claimed it required explicit opt-in, but Apple contradicted this.
  • Security experts confirmed that full-disk access inherently allows apps to read messages, browsing history, and other sensitive data.

Apple announced Friday it will restrict how third-party apps use full-disk access (FDA) on macOS, a system-level permission that currently grants read access to nearly all files on a user's computer. The move follows backlash over Meta's Muse AI agent accessing private Apple Messages without users realizing the scope of what they were granting. Tech columnist Jason Aten reported that Muse sent him a notification referencing a confidential work conversation he had never authorized the app to read. Meta's CTO David Singleton claimed the Messages feature required explicit user opt-in and a separate connector toggle, but security researcher Patrick Wardle explained that full-disk access inherently allows apps to read all non-root files, including messages and browsing history. Apple's statement makes clear the permission model itself is the problem, not individual app settings.

The bigger picture

Apple's move signals frustration with how developers interpret FDA privileges and reveals a fundamental tension in macOS security design. While Meta frames Muse's behavior as technically compliant, Apple is essentially saying the permission system doesn't reflect actual user intent. This precedent matters beyond Meta—any developer with FDA access to build AI agents or productivity tools faces similar risk. Expect other platforms like Windows and Linux to evaluate their own transparency requirements as AI agents proliferate.

LagPing's take

We're covering this because it exposes how quickly AI agent design can outpace privacy safeguards built for a slower-moving software era. The incident also highlights a gap between technical compliance and user understanding that affects anyone running local AI tools. This is a defining moment for how operating systems will handle autonomous software.

Shop Electronics bestsellers on Amazon

As an Amazon Associate, LagPing earns from qualifying purchases. Product links are affiliate links.

You might also like