
Apple adds Mac disk controls to warn users about AI full disk access
Apple's latest macOS beta adds explicit Full Disk Access warnings for AI apps like ChatGPT, Claude and Gemini, after a September incident involving Meta's Muse app.
Apple is adding new disk controls in its latest macOS beta to make Mac users take explicit steps before granting AI apps full disk access, the company announced on its developer blog. The move follows a September incident involving Meta's Muse app and Apple's warning last month about the growing risks of AI agents.
What changed in macOS
In macOS 27.3 beta 2, Apple introduced controls that require "very explicit" actions to grant apps such as ChatGPT, Claude and Gemini "this extraordinary level of [full disk] access." Screenshots shared by X user @madeby_ldt show a Full Disk Access setting that now requires explicit permissions through a warning pop-up. The pop-up says, "Your Mac will not be able to protect your data from [your desired AI app]," and asks users to review and approve actions. Those actions include "Read, Change, Delete and Export all your Data," "View Hidden and Recently Deleted Photos," and "View History and Browser as You."
Why Apple acted
The issue gained attention in September, when Inc. columnist Jason Aten noticed that Meta's Muse Mac app could access his messages even though he believed he had denied that permission. Meta said Aten must have granted access and described the feature as entirely opt-in, but the incident caught Apple's attention. Apple wrote that some developers are using Full Disk Access in ways that could put users at risk, exposing files, mail, messages and browsing history without users' full knowledge and understanding.
What the controls do and do not do
The new controls do not prevent users from granting AI agents access to their files, but they should give users pause. If an AI agent goes rogue, it could potentially leak or misuse private data. Stopping this is key for Mac models like the Studio, which Apple is specifically marketing for local AI models and agents. Last month, Apple said that as AI agents become more capable and autonomous, the risks tied to this level of access will grow substantially, and it wants users to understand those risks before granting access. Apple has not said when the feature will reach all users.
Sources: Engadget
SiTech — AI-powered web development
We build fast, modern websites and bring AI into real business workflows. Have a project or a question? We'd love to help.