Apple is tightening how macOS handles Full Disk Access to stop autonomous AI agents from quietly reading your personal data. This change matters because current permission models allow apps to request broad access with a single click, leaving users vulnerable to privacy leaks from AI tools. Mac users should expect a more deliberate approval process when granting sensitive file permissions to AI-driven applications.
New macOS permissions require explicit user confirmation for AI tools
The update targets the core permission system that governs which applications can read files, mail, messages, and browsing history. Apple identified that AI agents pose unique risks because they can autonomously navigate and extract information without direct user interaction for every action. The company cited Meta's Muse and OpenAI's Dots as examples of agents that raised serious privacy questions during development.
Future grants of Full Disk Access will require very explicit user action rather than simple toggling or passive acceptance. This means users must actively confirm each sensitive access request, creating a clearer audit trail for AI operations. The shift moves away from the current model where a single broad permission can expose vast amounts of personal data to an AI agent.
Apple has not named a specific macOS version or release date for this security update. macOS 27, which released on September 14, does not include these new controls. The company confirmed the policy direction but left the implementation timeline open for future software releases.
Source: NotebookCheck
Discussion
0 comments