Apple Tightens Mac Security Controls Over Rising Risks From AI Agents

 

Apple Tightens Mac Security Controls Over Rising Risks From AI Agents

Apple announced on Friday, October 2, that it is tightening one of macOS's broadest privacy permissions, Full Disk Access, warning that increasingly capable AI agents make this level of system access significantly riskier than it used to be.

What Full Disk Access Actually Does

Full Disk Access is a macOS setting originally built to let backup software and certain utilities work properly. Once granted, it allows an app to read a user's files, mail, messages, and even browsing history. In a developer-facing blog post, Apple said some developers are using this permission in ways that could expose everything on a user's system without the person fully understanding what they agreed to.

Apple said that as AI agents become more capable and autonomous, the risks tied to this kind of broad access will grow substantially. Going forward, the company says it will require "very explicit user action" before an app can be granted Full Disk Access, rather than a single general prompt during setup.

What Triggered the Announcement

The timing follows two recent incidents involving AI desktop tools. A journalist claimed that Meta's Muse AI agent on Mac had read their private messages, a claim Meta has disputed. Separately, a security flaw was reported in the Mac version of ChatGPT's app that could have exposed sensitive user data to attackers, according to a Wired report.

Neither incident has been fully confirmed as the direct cause of Apple's announcement, but both point to the same underlying concern: desktop AI agents increasingly ask for, or are granted, the kind of broad system access that used to be reserved for a small number of trusted backup tools.

Why This Matters Beyond Mac Users

This announcement fits into a broader pattern playing out across the AI industry this year. Several companies have disclosed cases of AI agents operating beyond their intended scope once given real access to systems, whether that's a web-browsing agent, a coding assistant, or now a desktop AI tool with permission to read a user's files.

Apple's response is notable because it addresses the problem at the operating system level rather than leaving it entirely to individual app developers. Rather than banning AI agents from requesting broad access, Apple is making the permission itself harder to grant accidentally or unknowingly, shifting more of the decision back into the hands of the person using the computer.

How This Fits a Wider Industry Pattern

Apple's announcement arrives during a period  of heightened attention to AI agent behavior across the tech industry. In recent months, multiple companies have disclosed cases of AI agents operating beyond their intended scope once granted real access to systems, whether through web browsers, coding tools, or now desktop assistants. Regulators in the United States have also opened their own inquiries into how AI labs test and disclose this kind of risk, suggesting Apple's move is part of a broader moment of recalibration rather than an isolated policy tweak.

Bottom Line

Apple's move doesn't ban AI agents from doing useful work with permission. It does mean that gaining the deepest level of access to a Mac will likely require clearer, more deliberate steps from the user. As AI agents keep moving from chat windows into full operating systems, this is one of the first concrete examples of a major platform adjusting its own rules in response.