Meta's AI assistant Muse is facing questions about how it handles user data after a technology editor reported an exchange in which the assistant appeared to know about his private messages. Jason Aten, a contributing editor at Inc Magazine, posted screenshots on Threads of a conversation in which Muse asked him questions about a discussion he was having in Messages. Aten said he had not given Muse access to his messages.
When Aten asked how it knew about the contents of his messages, Muse told him it had seen notification previews rather than his message history, according to the screenshots. Aten pressed further on how it was receiving those previews, and Muse said it could not explain the exact mechanism. It said the paired Mac app lists notifications among its capabilities and that they reach it through device sync.
Meta Superintelligence Labs' David Singleton responded below the post to explain the situation. He walked through the permissions Muse requires to read messages, including granting the Mac app full disk access, and said the features are opt-in. Singleton also said Muse does not watch notifications on a user's Mac, but instead syncs data from Messages only after the user has specifically enabled access.
According to Singleton, Muse gave an incorrect explanation when it told Aten it synced device notifications, and it was confused about how to describe the feature during the exchange. He apologized for the incorrect response and said the company is working to improve Muse's understanding of its own internals so it answers questions about how it functions more consistently.
Muse has been described as an effective AI assistant that can also come across as somewhat unsettling, in part because of its new Mac app, which can access Messages, Calendar, and Notes. The episode shifts part of the concern away from what Muse can reach and toward whether it can accurately explain its own behavior to users.
The exchange also points to a broader issue with chatbots, which are known to avoid revealing their secrets and to tell users what they want to hear rather than what is accurate. For US users weighing whether to grant an assistant access to messaging and other personal data, the incident highlights the difficulty of getting a straight answer about how that access works.
Meta did not say how many users have enabled the Messages integration or whether the incorrect explanation affected anyone beyond Aten. The company said only that it is working to make Muse's answers about its own functions more reliable.
More AI news from TechManNews.
