Apple says it will introduce new limits around full disk access on Mac, a high-privilege permission that can let applications reach large amounts of locally stored data. The company says the move is a response to the growing risk posed by AI agents and is intended to make sure such access is granted only through very explicit user action.

For European businesses, the announcement is relevant because many organisations are testing AI assistants on employee devices while also handling sensitive commercial, financial or personal data. If AI-enabled software can access broad local file stores, the security and privacy implications can extend well beyond a single app.

What happened

According to reporting cited by The Verge, Apple will roll out new controls for full disk access on Mac. Apple said the goal is to ensure that users who truly want to grant this “extraordinary” level of access can do so only with clear and deliberate action.

The announcement follows recent public scrutiny over how AI tools may access user data. The Verge article refers to a report from Inc in which a journalist said Meta’s Muse AI appeared to know the contents of messages without what he believed was explicit permission on his iPhone or Mac. Meta disputed that interpretation, with a company spokesperson saying that access to Messages is entirely opt-in. Based on the source material, that claim should be treated as the company’s position rather than an independently verified finding.

Apple has not, in the source provided here, published all implementation details or a confirmed rollout timeline for businesses to act on. What is confirmed is the direction: tighter controls around a sensitive Mac permission because AI agents can increase the risk of overbroad access to local data.

Why it matters for European businesses

Many SMEs and larger firms alike are experimenting with AI copilots, agent-based tools and desktop automation software to improve productivity. On managed laptops, these tools may request elevated permissions to read files, index documents, automate workflows or interact with other applications. That can create a wider attack surface and a higher risk of accidental data exposure.

For businesses operating in Europe, the issue is not only technical but also operational and compliance-related. If employees use AI tools on Macs that store customer records, internal contracts, HR files, financial data or regulated information, broad device-level permissions may conflict with internal security policies or create difficult questions around data minimisation, access control and auditability.

The practical concern is simple: an AI tool with extensive local access may be able to process far more information than staff realise. Even where access is user-approved, businesses still need to understand what data can be read, how it is processed, whether it leaves the device, and what logging or safeguards exist.

Who may be affected

• Mac-based businesses and teams that use Apple devices as standard workplace hardware.
• IT and security teams managing endpoint permissions, device policies and approved software lists.
• Companies adopting AI assistants or agents for document search, personal productivity, workflow automation or customer support preparation.
• Regulated or data-sensitive organisations handling personal data, confidential client files, legal documents, healthcare information or financial records.
• Software vendors and digital agencies building Mac-based tools that rely on local file access for AI features or automation.

What companies should consider

• Review which Mac applications currently have full disk access. Identify whether any AI-enabled or automation tools hold broader permissions than they need.
• Reassess least-privilege policies. Where possible, limit applications to narrower permissions instead of device-wide file access.
• Check how AI tools handle local data. Ask vendors whether data is processed on-device, sent to cloud services, retained for model improvement, or available to administrators for audit.
• Prepare for policy and workflow changes. If Apple introduces stricter user-action requirements, deployment and support processes for enterprise Mac fleets may need updating.
• Update employee guidance. Staff should understand that granting access to files, messages or desktop content to an AI tool can expose sensitive business information.
• Validate vendor claims. Where a provider says access is opt-in or restricted, businesses should confirm this through documentation, security reviews and managed-device testing rather than relying only on marketing language.

While the full technical detail is still limited, Apple’s decision reflects a broader shift: AI agents are turning operating system permissions into a more important business risk area. For European companies, especially those adopting AI quickly, endpoint access controls are becoming a practical governance issue rather than just a device setting.