Apple disclosed plans to modify its macOS privacy framework in response to growing concerns about how developers are utilizing full-disk access permissions. The company stated that some developers have been accessing full-disk access in ways that could expose users to privacy risks, potentially granting unrestricted access to files, emails, messages, and browsing histories without adequate user awareness.
The announcement followed a public incident involving Meta’s Muse AI assistant, which a tech columnist reported had accessed his private Apple Messages communications. Meta’s chief technology officer claimed the assistant could only read messages if users explicitly enabled both a system-level full-disk access permission and a Messages connector setting within the application. Security researchers subsequently challenged this assertion, arguing that full-disk access inherently grants applications the ability to read all unencrypted files on a system, including message content, regardless of additional settings.
Apple emphasized that as AI agents become increasingly autonomous and capable, the risks associated with granting such extensive system access are expected to increase substantially. The company stated it would work to ensure users have clearer understanding of privacy implications before authorizing full-disk access permissions, allowing individuals to make more informed decisions about their data.
The timing of Apple’s announcement coincided with broader security concerns about Muse. A macOS security researcher had previously disclosed a configuration vulnerability in the assistant that could allow other applications or malicious code to assume control of Muse and access its granted resources. Additionally, Amazon had blocked Muse from its platform, stating that AI applications of this type should operate transparently and respect the decisions of service providers regarding participation.
Meta did not provide additional comment beyond reiterating its previous statements about the opt-in nature of the Messages integration. Industry observers suggested that the combination of these developments indicated potential risks associated with granting such comprehensive system-level access to AI assistants.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI