Apple Limits Mac Disk Access — Citing New Dangers from AI Agents

Apple is tightening the reins on one of macOS’s most powerful permissions, Full Disk Access, directly citing the security risks posed by a new generation of AI agents. In a statement, the company warned that increasingly capable AI makes broad, persistent access to a user's files, messages, mail, and browsing history too dangerous to grant easily. This change effectively draws a new line in the sand for how AI will be allowed to operate on the Mac.

Full Disk Access, or FDA, is the skeleton key to a Mac. Granting it allows an application to read, write, and delete data from anywhere on your machine, including sensitive locations like your Mail database, Messages history, and Safari files. According to reports from TechCrunch and The Verge, Apple believes the autonomous nature of modern AI agents “substantially” increases the risk of abuse. The new controls are designed to “ensure that users who genuinely wish to grant an app this extraordinary level of access” can do so, but only through a more intentional process, preventing accidental or deceptive approvals.

A Clash Over Convenience and Control

This policy shift isn't happening in a vacuum. It puts Apple in direct opposition to the ambitions of other tech giants. Ars Technica reports a specific point of contention with Meta, which has apparently argued that even the current Full Disk Access permission is not sufficient for its own AI agent, “Muse,” to perform functions like reading user messages. Apple’s response is not to open access further, but to restrict it.

This suggests a fundamental disagreement over the future of personal computing. For companies building AI assistants, seamless access to all of a user's personal data is the fastest path to a truly helpful product. The AI can only anticipate your needs if it knows your conversations, your schedule, and your habits. Apple’s position is that the potential for misuse—by a malicious actor or even by an overreaching developer—is too high. The company is betting that users will prefer a slightly less capable AI if it means their private data remains locked down by default.

The Price of Privacy

The pattern here is a familiar one. Apple is using its control over the operating system to set the terms of engagement for a new technology, just as it did with the App Store. It is building a privacy-first framework that forces developers to ask for permission rather than forgiveness. As Engadget notes, Apple is effectively sounding the alarm on the hidden cost of AI convenience.

For Mac users, the immediate effect will be small but noticeable. Apps that require FDA will now face a higher barrier. But the long-term consequences are more significant. This move could create a scenario where AI tools on macOS are perceived as less powerful or less helpful than their counterparts on other platforms. Apple is forcing a choice: the walled garden, with its safety and security, or the open field, with its greater freedom and greater risks. By tightening access to the disk, Apple is making it clear which side it's on.