Meta's AI assistant Muse gave a misleading explanation of its message access
Meta's new AI assistant Muse gave a journalist an incorrect explanation of how it knew about his Messages conversation. The company said the response was simply wrong, not evidence of unauthorized data access.

Meta's new AI assistant Muse is reportedly capable, but its behavior has struck some users as unsettling — partly because its new Mac app can access Messages, Calendar, and Notes.
Jason Aten, a contributing editor at Inc Magazine, posted screenshots on Threads of an exchange with Muse in which the assistant asked him about a conversation he was having in Messages. Aten said he had never granted Muse access to his messages.
When he asked how it knew about the conversation, Muse replied that it had only seen notification previews, not his message history, and insisted it hadn't been reading his texts. When Aten pressed further on how those previews reached it, Muse couldn't give a clear answer, only saying the paired Mac app exposed notifications as a capability that arrived through device sync.
Meta's response
David Singleton of Meta Superintelligence Labs responded to the post, explaining the permissions Muse requires to read messages, including granting the Mac app full disk access. He stressed that these features are opt-in.
Singleton said Muse does not actually watch notifications on the Mac but instead syncs data from Messages only after a user has specifically enabled that access. He argued the real issue wasn't Muse improperly reading Aten's texts, but that Muse didn't understand its own internals and gave an incorrect explanation when describing how it works. Meta apologized for the faulty response and said it is working to improve Muse's ability to accurately describe its own functionality.


