Apple says it will add controls around macOS Full Disk Access as increasingly autonomous AI agents seek access to personal information. TechCrunch reports that the company wants users to make a more explicit decision before giving an application this broad permission. The announcement follows a disputed account of Meta’s Muse reading private messages and a separate report about a vulnerability in ChatGPT’s Mac application.
The change shifts attention from what an assistant can do to what a user has authorised it to reach. An agent that works across a computer may need information from several places to complete a task. Apple’s statement argues that the breadth of Full Disk Access can expose more than a user understands, making the consent process itself part of the security problem.
A broad permission faces new scrutiny
Full Disk Access was designed to help backups function properly, according to Apple’s explanation cited by TechCrunch. It can give an application access to files, mail, messages and browsing history. Those are different kinds of personal information, brought within reach through a setting whose consequences may be difficult for a user to judge simply by looking at an assistant’s immediate task.
Apple says some developers are using that access in ways that could expose everything on a system without sufficient user awareness. The company attributes the increased risk to AI agents and expects it to grow as they become more capable and autonomous. This is Apple’s assessment of the permission’s changing use, rather than a finding that every agent granted access has behaved improperly.
Its planned response is to require very explicit user action when someone wants to grant such extensive access. Apple also says users should understand the risks well enough to make an informed choice about their data. The report does not describe the detailed interface, implementation or release schedule of the new controls. The announcement therefore establishes a direction for macOS permissions, not a completed change users can already evaluate.
The Muse allegation remains disputed
The timing matters because Inc. columnist Jason Aten recently said Muse knew the contents of private messages despite his belief that he had not granted it permission. Meta disputed that account. TechCrunch presents the incident as an allegation and a denial, and Apple’s announcement does not resolve the disagreement. It should not be read as confirmation that Muse bypassed a required setting.
In Muse, Full Disk Access is an optional permission that a user can enable. The surrounding concern is about desktop assistants that can read files and messages or control activity on a computer. The report connects that capability to questions of trust: a person may welcome help with a task while still expecting a clear account of which information the assistant can obtain.
TechCrunch also notes a WIRED report describing a ChatGPT Mac app flaw that could have allowed hackers to access sensitive data. That is a separate security issue from the Muse permission dispute. One concerns a reported software vulnerability; the other concerns competing accounts of an assistant’s access to messages. Their inclusion in the article explains the environment in which Apple made its statement without proving that both incidents had the same cause.
Clearer consent still needs implementation details
Apple’s proposal focuses on the decision to grant broad access. It leaves open how the company will communicate the scope of that permission and what exact steps will be required. TechCrunch says Apple did not respond to its inquiry about the feature change. Without those details, the report cannot establish how the new process will affect individual applications or their users.
More explicit consent also answers a specific question: whether a user knowingly chose to give an application extensive reach. It does not, by itself, explain what happened in the disputed Muse case or demonstrate that every permitted application’s software is secure. Those questions would require evidence beyond the consent policy Apple has announced.
For now, the development is a stated commitment to tighten access controls as AI agents expand their role on desktops. Apple’s rationale is that capability and autonomy increase the consequences of broad permissions. The practical effect will depend on the controls it delivers and how clearly those controls show users the data access they are choosing.
