Apple is changing how macOS handles requests for access to sensitive Mac data as AI agents become more capable of performing tasks on behalf of users.
The company plans to add stronger warnings and require more explicit user approval when applications request Full Disk Access, a macOS permission that can give an app access to a broad range of files and information stored on a computer.
The change comes after complaints surrounding Meta’s Muse AI agent, which has faced questions about how it accesses information on users’ Macs.
The controversy highlights a growing privacy issue. AI agents can now do more than answer questions. They can interact with applications, search for information, manage tasks and use data stored on a user’s device.
To perform these tasks, some agents may request access to large amounts of information. Apple wants users to have a clearer understanding of what they are allowing before granting that level of access.
ALSO READ: Meta’s New Muse AI Agent Can Send Emails, Book Travel and Make Payments
Apple Plans Stronger Warnings for macOS Full Disk Access
Apple is planning changes to Full Disk Access, a macOS permission that can allow an application to access a wide range of files and data stored on a Mac. The permission has legitimate uses. Backup and security software, for example, may need broad access to files and system information to perform certain functions.
However, Apple said the feature can also create privacy risks when applications request more access than they need. Under the planned changes, Mac users will have to take a more deliberate action before giving an application Full Disk Access.
Apple wants the new process to make the potential risks clearer so users can better understand what they are approving. Apple also pointed to the growing capabilities of AI agents as a reason for strengthening these controls.
“As AI agents become increasingly capable and autonomous,” Apple said, broad access to a computer could create greater risks. An agent with extensive permissions could potentially interact with files, applications and other sensitive information without requiring constant user input.
Why Apple Is Rethinking Mac Permissions for AI Agents
Traditional applications are generally built to perform specific tasks and often have limited access to a user’s data. AI agents are different because they can handle multiple steps and work across different applications with less user involvement.
An AI agent with broad system permissions could potentially access files, emails, messages, browsing history and other sensitive information stored on a computer. The more tasks an agent can perform on its own, the more important those permissions become.
This creates a different privacy risk. A user may grant an AI agent access to complete a particular task without fully realizing how much additional information that permission could expose.
Apple’s planned changes are intended to make the scope of Full Disk Access clearer before users approve it. The company wants users to have a better understanding of what they are allowing an application or AI agent to access.
Meta’s Muse Raises Questions About AI Agent Access to Private Messages
The planned changes from Apple come after a recent controversy involving Meta’s Muse, an AI agent that can perform tasks on behalf of users. Muse can handle tasks such as managing subscriptions and searching for better prices.
However, technology columnist Jason Aten recently claimed that Muse referenced content from private conversations in Apple Messages on his Mac, despite saying that he had not given the agent permission to access those messages.
The claim raised questions about how AI agents interact with personal data stored on computers and how much access they may have once users grant broad system permissions. Meta disputed the claim that Muse could access Messages without the necessary permissions. Meta spokesperson Andy Stone said the Messages integration in the Mac version of Muse is opt-in.
According to Meta, users must enable both Full Disk Access and the Messages connector before Muse can access Messages content. Meta also said users can revoke these permissions at any time.
The dispute has drawn attention to a broader issue: users may not always understand how much data an AI agent can potentially access when they grant it broad permissions at the operating-system level.
ALSO READ: Meta’s Muse AI Phone Calls Involved Human Contractors
How Mac App Permissions Differ From iPhone Protections
Apple already uses stricter app isolation on iPhones and iPads. Through sandboxing, apps generally cannot access data belonging to other apps unless Apple provides a specific system feature that allows it.
macOS gives applications more flexibility because some software needs broader access to files and system resources. Full Disk Access is one example of this approach. That flexibility is useful for applications such as backup and security tools, which may need access to large amounts of data to perform their functions.
However, it also means an application that receives user approval can potentially access much more information than a typical iPhone or iPad app. Apple now wants Mac users to have a clearer understanding of this trade-off. The change becomes more important as AI agents gain the ability to perform more tasks with less direct user involvement.
Apple’s New Permission Rules Could Change How AI Agents Access Macs
Apple’s planned changes could make privacy permissions a more important consideration for companies developing AI agents for desktop computers. AI companies are working to make agents capable of completing more tasks with less input from users. To do that, agents may need access to multiple applications, files and other sources of personal information.
However, broader access also increases the potential impact of mistakes, unexpected actions or unclear user consent. An agent with extensive permissions could interact with information that goes beyond what a user originally intended to share.
Apple’s changes do not prevent AI agents from receiving Full Disk Access. Instead, the company is adding clearer warnings and requiring users to take a more deliberate action before granting the permission.
The change could also encourage developers to request more limited permissions where possible instead of relying on broad system-level access.
What Apple’s Changes Mean for the Future of Mac AI Agents
Apple has not yet provided a detailed timeline for all of the planned changes to the macOS permission system. The company has said it will introduce additional controls that require users to take a more explicit action before granting Full Disk Access.
The changes come as AI agents are becoming capable of working across multiple applications and completing tasks on behalf of users. As these systems gain more control over computers, the permissions they receive are becoming an increasingly important part of their design.
For Apple, the challenge will be to support these new capabilities while maintaining its focus on user privacy. For AI developers, the controversy surrounding Muse and Apple’s response could also lead to greater attention on how permissions are requested, explained and tested.
As AI agents gain more control over a user’s computer, users need a clear understanding of what those agents can access and what they can do with that access.
