macOS: Apple Tightens Full Disk Access – AI Agents as the Trigger

TVS News AI, editorially supervised, presents macOS: Apple Tightens Full Disk Access – AI Agents as the Trigger.

When AI agents act autonomously through the system, they often need far-reaching permissions – and that is exactly what has now put Apple on alert. The company is apparently planning stricter controls for Full Disk Access under macOS. Apple cites growing security risks from autonomous AI assistants that, on behalf of the user, read files, send emails, or change system settings.

Full Disk Access has been one of the most powerful tools on the Mac for years: programs with this permission can access user data, documents, messages, and system files with virtually no restrictions. What was originally intended for backup tools, antivirus programs, or disk utilities, however, also opens the door wide for software that behaves maliciously.

Apple plans stricter controls for Full Disk Access. According to Apple, the reason is growing risks from AI agents.

Why Apple is tightening the rules now

According to Golem.de, Apple sees new risks through AI agents that go beyond classic malware. Autonomous software that carries out tasks on behalf of the user gains particularly deep system access through Full Disk Access. It is precisely this combination of far-reaching permissions and automated actions that Apple wants to rein in going forward.

Concretely, this means: an AI that autonomously sorts emails, rewrites documents, or executes code effectively has free rein on the Mac with Full Disk Access. That is convenient – but also a godsend for any attacker who compromises such an app.

What Full Disk Access means today

Anyone who grants an app on the Mac full disk access allows it to access sensitive areas such as mail databases, Safari history, messages, notes, documents, and system logs. macOS does actively prompt users to grant the permission only to trusted apps – but in practice, many click through the prompt without being aware of its implications.

This is exactly where Apple's considerations come in. The company apparently wants to prevent AI agents from becoming a master key to the entire system with a single click.

What users can expect

It remains unclear how concretely Apple will tighten the controls. Stricter warnings when activating for the first time, time limits, a requirement for regular confirmation – similar to microphone or camera access – or more detailed logging of what an authorized app actually does are all conceivable. However, all of this is currently speculation.

What users can already do now: review the permissions they have granted. Under System Settings > Privacy & Security > Full Disk Access, you can check at any time which apps currently have full access – and if in doubt, you can revoke it there as well.

What remains unclear

The report so far leaves several central questions open. It is unclear, among other things, from which macOS version the new rules will take effect, whether existing permissions will be automatically revoked, and how Apple intends to distinguish between legitimate AI tools and potentially dangerous software. The question of whether third-party providers will need to actively follow suit also currently remains unanswered.

As soon as Apple announces concrete steps, it will likely also become clear how serious the company is about protection – and how much convenience Mac users will have to give up for it.