Apple is tightening the 'Full Disk Access' permission controls in macOS to mitigate privacy risks associated with AI agents. The company plans to introduce new security mechanisms that require users to perform more explicit actions when granting high-level access to applications.
The initiative follows concerns regarding how AI agents, which can autonomously perform tasks across files, emails, and messages, utilize system permissions. Apple noted that some developers are requesting access in ways that may leave users unaware of the extent of data exposure.
Recent controversies, including discussions around Meta's Muse AI agent and security reports involving the Mac version of ChatGPT, have highlighted the risks of AI agents accessing sensitive user data. Apple is re-evaluating its security architecture to address these evolving threats.
Apple previously introduced a new security framework at WWDC for Xcode and AI coding agents, which restricts access to project files based on system-level policies. The company aims to balance the utility of autonomous AI agents with the necessity of protecting personal digital environments.
