Apple is changing how macOS handles Full Disk Access as it responds to a new class of risk tied to AI agents. The company says increasingly capable agents make it more dangerous for apps to gain broad access to users’ files and personal data, including messages, mail, and browsing history.
Why Apple is changing the permission model
Full Disk Access has long been one of macOS’s most powerful permissions, letting an app read large portions of a system’s data. Apple’s concern is that AI agents can now make use of that access in more unexpected ways, turning a convenience feature into a wider privacy exposure. In the reports supplied here, Apple says it will add new controls to address that risk and ensure that users who truly want to grant that level of access do so through very explicit action.
The timing of the change follows recent attention on an AI-related privacy issue. One report described Meta’s Muse AI as apparently able to learn the contents of messages despite the user not giving explicit permission to access them on iPhone or Mac. That episode helped sharpen the discussion around what broad system permissions mean when AI tools are involved.
What Apple says will change on Mac
According to the supplied coverage, Apple is not removing Full Disk Access entirely. Instead, it is introducing new limits and controls around the permission so that granting it requires a more deliberate user step. The goal is to make it harder for an app to quietly receive a level of access that can expose far more information than a user may realize at the moment of approval.
This shift matters because the permission can reach into sensitive areas of daily computing. If an app can inspect files, messages, mail, and browsing history, then even a helpful AI agent can become a much broader data reader than the user intended. Apple’s response appears aimed at reducing that gap between perceived and actual access.
How AI agents change the security equation
The supplied articles frame AI agents as different from traditional apps because they may act across a system in more dynamic ways. An ordinary application might request access for a narrow task, but an agent can use broad permission to gather and process data across many personal contexts. That makes the risk less about a single app feature and more about what happens when an autonomous or semi-autonomous system is given a deep view into the machine.
That is also why Apple’s update is being described as a response to the “substantially” increased risk posed by AI agents. The concern is not simply that a program can read data, but that newer AI tools can make broad system permissions more valuable, and therefore more dangerous, than before. Apple’s new controls are meant to keep users in the loop before such access is granted.
What users should watch for
For Mac users, the practical takeaway is that permission prompts around Full Disk Access may become more explicit. The supplied reporting indicates Apple wants to ensure that only users who genuinely wish to grant this extraordinary level of access can do so. In other words, the company is trying to make the approval process more clearly intentional.
That does not eliminate the need for caution. A permission that can expose messages, mail, browsing history, and files still carries serious privacy implications, even if it is granted through a better prompt. Users will still need to decide carefully which apps and agents should receive such access, and why.
The bigger picture for platform security
Apple’s move reflects a broader tension in consumer software: tools built to be helpful often work best when they can see more of what a user is doing, but that visibility can also increase privacy risk. AI agents amplify that tension because they may need broad context to be useful. The result is a pressure on operating systems to rethink old permission models rather than simply applying them unchanged to new kinds of software.
In that sense, the Full Disk Access change is less about one app or one incident than about adapting to a changing ecosystem. As AI tools become more capable, platform vendors are being pushed to decide how much access should be possible, and what form consent should take when that access reaches deeply into personal data.
Conclusion
Apple’s decision to tighten macOS Full Disk Access controls shows how AI agents are already reshaping basic security assumptions. By requiring more explicit user action before granting such broad access, Apple is trying to reduce the chance that powerful new software can quietly see too much. The change underscores a simple point: as AI agents become more capable, the permissions that support them need to become more careful too.