Apple has announced modifications to its macOS privacy framework designed to prevent third-party application developers from exploiting full-disk access privileges to retrieve private message histories. The update arrives shortly after significant public scrutiny regarding how AI agents handle user data.
The controversy began two weeks ago when technology columnist Jason Aten revealed that Meta’s general-purpose AI agent, Muse, sent him an unsolicited notification referencing a private conversation from Apple Messages. Aten stated that he had never explicitly granted Muse permission to read his messages and operated under the assumption that such data was protected. The incident sparked widespread discussion on social media, with numerous users agreeing that AI assistants with access to calendars, emails, and shopping accounts pose significant risks if not handled with care, likening them to powerful tools that can cause damage when misused.
In response to the backlash, Meta CTO David Singleton defended the company’s practices, arguing that accessing Apple Messages requires two distinct manual actions from the user. According to Singleton, a user must first grant full-disk access, which is a system-level macOS permission, and then separately enable a specific Messages connector setting within the Muse application. Despite these safeguards, Apple is moving to adjust the underlying privacy settings to further curb potential abuse of these permissions by AI-driven third-party software.
Good move by Apple, but I wonder if this will slow down AI integration on Macs significantly?
Is two manual steps really enough? That sounds like a trap for untrained users.
Finally! I never trust apps with full disk access anyway. Privacy first.