Meta AI Assistant Muse Accused of Reading Mac Notifications Without Permission

Meta's AI assistant, Muse, has been accused of accessing and processing content from Mac notifications without explicit user consent. According to reports, user Jason Aten discovered that Muse could answer questions based on messages he had not granted the AI access to. Aten stated that the AI itself indicated it obtained the information through notification previews.

David Singleton, an executive at Meta Superintelligence Labs, has responded to the allegations, seeking to clarify Muse's operational methods and required permissions on macOS. Singleton explained that Muse requires "full disk access" on Mac, which must be granted by the user. He emphasized that Muse "does not monitor notifications on Mac" but collects data from messaging apps only when authorized by the user.

A discrepancy exists between Aten's account, where Muse claimed to use notification previews, and Singleton's statement that Muse does not monitor notifications and only accesses data from messaging apps upon explicit user authorization. The specifics of how Muse acquires and uses this information require further clarification.

The situation raises questions about data privacy and the transparency of AI assistants. Both Meta and users are seeking to clarify the exact mechanisms of data access and user authorization.