Apple Strengthens macOS Security Controls to Restrict AI Agents’ Disk Access

image 844
🔊 Listen to the summary

Apple has introduced new security controls in its latest macOS beta that require users to take explicit steps before granting AI agents full access to their storage drives. The update targets software like ChatGPT, Claude, and Gemini to ensure individuals realize the extent of data they are exposing to automated tools.

Under the revised Full Disk Access settings, macOS presents an explicit warning interface before permissions can be finalized. The prompt cautions that the system will no longer be able to protect data from the specified application, listing actions the tool will gain authorization to perform. These permissions include reading, altering, deleting, and exporting all user data, inspecting browsing histories, and browsing hidden or recently deleted photos.

Scrutiny over AI access heightened in September when Inc. columnist Jason Aten discovered that Meta’s Muse application accessed his personal messages, despite his belief that he had declined permission. While Meta maintained that the feature was entirely opt-in and required user authorization, Apple stated that certain developers were utilizing Full Disk Access in ways that expose files, emails, messages, and browsing histories without sufficient user understanding.

Apple previously noted that growing autonomy in AI agents will escalate the hazards tied to broad storage access. While the new measures do not block users from granting permissions entirely, they force users to review the potential exposures before confirming. The updated safeguards are currently available in macOS 27.3 beta 2, with Apple yet to announce a public release date.

Source: engadget.com