Apple is adding stricter controls to a powerful macOS feature, responding to privacy risks from AI agents.
Outlets are counted by registrable domain, so a broadcaster’s station subdomains count once. Every item here is from a different masthead.
Apple is adding stricter controls to a powerful macOS feature, responding to privacy risks from AI agents.
Apple announced it will introduce new controls for a macOS setting called Full Disk Access, requiring 'very explicit user action' for an app to gain that permission. The company said some developers are misusing the setting, which was originally designed to let backup apps work properly.
Apple stated the feature can expose files, mail, messages, and browsing history without a user's full understanding. The company linked the change directly to the rise of AI, saying 'as AI agents become increasingly capable and autonomous, the risks... will grow substantially.'.
The announcement follows a report from Inc. columnist Jason Aten, who said Meta's Muse AI app knew the content of his private Apple Messages, though he claimed he had not given it permission.
Meta disputed Aten's account. A company spokesperson said access to Messages is 'entirely opt-in,' requiring a user to enable both Full Disk Access and a specific Messages connector within Muse. Despite this, the report raised widespread questions about AI agents having broad access to personal files and communications.
Ars Technica reported that a macOS security expert questioned Meta's defense, arguing that any app with Full Disk Access can technically read non-system files like messages. When pressed on this point, Meta's only response was to reiterate its statement that both permissions were required.
Apple's new blog post, aimed at developers, appears to side with the concerns about unclear disclosure. The company said it is changing Full Disk Access to ensure users clearly understand the risks before granting such sweeping access to their systems.
3 independent outlets filed 3 reports over 5 hours.
This story is being driven by new information and intent from Apple, according to the engine's analysis. The company's own directed intelligence—its ability to shape its ecosystem—is at a high level, driving the push for change in response to new risks.
A key factor is the intensity of exploration drive, which the engine measured as critically high. This means novel AI agent behavior is the primary destabilizing variable, creating a new threat that Apple feels compelled to address with its permissions system.
The cost and friction of implementing this change is also measured as high. That cost—the price of a more secure system—will be paid by developers and users, creating the potential for resistance. Apple's calculated move aims to direct the evolution of software toward a more secure equilibrium.
The engine's primary projection, with a 60% probability to be settled by October 2026, is that Apple successfully tightens the controls. This outcome would be confirmed when at least two independent sources report that the new permission system prevents AI agent abuse while maintaining legitimate workflows.
A secondary branch, given a 30% chance to be settled by October 2026, forecasts developer frustration. This would be confirmed by reports that overly restrictive controls prompt developers or users to create security-weakening workarounds, undermining Apple's intent.
A less probable but intense simulation, at 10% and judged by mid-October 2026, anticipates rapid AI agent evolution. It would be settled if reports show developers quickly adapting to create sophisticated methods that circumvent the new permission controls entirely.
The engine's analysis shows a strong probability Apple will succeed in raising the security floor for its macOS platform through this defensive move. However, it assigns a combined 40% probability to unintended consequences, either porous workarounds or a new arms race.
The outcome will be determined by the reaction in developer communities and AI agent ecosystems, which the engine identifies as the meso-level battleground. Watch for concrete data on developer pushback or agent adaptation as the next falsification event that will shift these probabilities.
Meta says FDA isn't sufficient to Muse reading messages. Apple begs to differ.
Apple will add new limits for "full disk access" on Mac in response to risks posed by AI agents, as reported earlier by TechCrunch. In an update on Friday, Apple says it's rolling out new controls to "ensure that users who genuinely wish to grant an app this extraordinary level of access can only do […]
Apple says it will add new controls around macOS’s Full Disk Access permission, warning that increasingly capable AI agents make broad access to users’ files, messages, mail, and browsing history riskier.
Sources are evidence, not content. Each keeps its own name, its own link and an extract capped at 400 characters; none of it is rewritten into the copy above.
