Apple is enhancing security measures around macOS’s Full Disk Access feature due to growing concerns about AI agents accessing sensitive user data. This announcement follows reports that Meta’s Muse app on Mac read users’ private messages without clear permission and a Wired investigation revealing vulnerabilities in ChatGPT’s Mac app that could expose user data to hackers. Full Disk Access allows apps broad permissions to view files, emails, messages, and browsing histories, which can create significant privacy risks when granted without careful consideration.

The controversy began when Inc. columnist Jason Aten reported that the Muse AI agent had access to his personal messages, despite him not explicitly granting the app permission. Meta has disputed this claim, but the incident amplified public fears about the transparency and security of AI-driven desktop applications. Apple’s statement acknowledges that some developers exploit Full Disk Access in ways that users may not fully understand, potentially exposing private information without proper user awareness or consent.

To address these issues, Apple plans to roll out new controls that require more explicit user consent before granting apps Full Disk Access. Instead of limiting the permissions technically, the company intends to ensure that users are fully informed about the risks associated with this powerful level of access. Apple emphasized that as AI agents become more autonomous and sophisticated, controlling these permissions carefully is essential to safeguard user data and privacy.

While Apple did not provide specific details on the timeline or technical nature of the changes, the company highlighted its commitment to helping users make well-informed decisions about their privacy as AI capabilities expand. This move signals growing industry recognition of the potential security challenges posed by AI agents operating with broad system access on personal devices, underscoring the need for explicit and clear user control over sensitive permissions.