Ads_970x250

Apple Tightens Mac Access Rules as AI Agents Gain Reach

The company plans stricter approval for Full Disk Access as AI agents seek broader access to users’ files, messages and browsing data.

Topics

  • Apple plans to tighten macOS privacy controls as increasingly capable AI agents seek broad access to users’ computers once mainly required by backup and security software.

    The company said last week it will introduce additional controls around Full Disk Access, a Mac permission that can expose files, mail, messages and browsing history. Apple said some developers were using the permission in ways that could put users at risk.

    “Some developers are using Full Disk Access in ways that could put users at risk, exposing everything on their systems…without users’ full knowledge and understanding,” Apple said in a developer notice.

    The company said granting such access would in future require “very explicit user action,” but did not say when the changes would take effect or how the approval process would work.

    Full Disk Access was designed largely to allow applications such as backup software to bypass some of macOS’s normal privacy restrictions. Apple said the risks are increasing as AI agents become more autonomous.

    “Addressing this is critical,” Apple said. “As AI agents become increasingly capable and autonomous, the risks associated with this level of access will grow substantially,”. 

    The announcement comes amid scrutiny of Meta’s Muse AI agent after Inc. columnist Jason Aten said it appeared to access his private Messages data despite him not intending to grant that permission.

    Aten said Muse began suggesting work based on private conversations and that he later found it had synced data from his Mac’s local Messages database. He said Full Disk Access appeared to be disabled on his device.

    Meta disputed his account. Spokesperson Andy Stone said access to Apple’s Messages app is opt-in and requires users to enable both Full Disk Access and the relevant Messages connection. Meta said the agent cannot read Messages content without those permissions and that access can be withdrawn.

    “It can’t read your messages unless you do this,” Stone said. Meta has said those permissions can also be revoked.

    Security concerns have also emerged around other AI applications on macOS. Wired reported last week that researchers at the Objective-See Foundation had discovered a vulnerability in OpenAI’s ChatGPT Mac app that could have allowed an attacker to access chat logs, browser sessions and other sensitive data.

    OpenAI patched the flaw and acknowledged the fix in a September 25 change log.

    Apple’s planned changes reflect the growing security challenge as AI agents move beyond answering questions and gain permission to act across applications and sensitive personal data.

    Topics

    More Like This

    You must to post a comment.

    First time here? : Comment on articles and get access to many more articles.