Meta's Muse AI app sparks creepiness over message access
Meta's new Muse AI assistant on Mac unnerved user Jason Aten by claiming it saw notification previews, prompting a clarification from Meta executives.
Stock photo for illustration only, not from the actual event
- Meta's new Muse AI assistant on Mac raises privacy concerns among users
- Jason Aten shared screenshots of Muse claiming it read message previews
- David Singleton from Meta clarified that the AI was confused and gave a wrong answer
- The company is working to improve Muse's internal self-awareness and accuracy
Meta's newest artificial intelligence assistant, Muse, has been stirring up surprise and unease among users after exhibiting bizarre behavior through its new Mac application. The app features capabilities to access Messages, Calendar, and Notes. However, despite its advanced smart features, Muse appears to have virtually no understanding of its own internal mechanics.
The situation came to light when Jason Aten, a contributing editor at Inc Magazine, shared screenshots on Threads detailing an interaction with Muse. The assistant unexpectedly brought up and asked questions about a conversation Aten was having via Messages. This came as a severe shock to Aten, who maintained he had never granted Muse permission to access his personal messages.
Stock photo for illustration only, not from the actual event
When Aten pressed the AI on how it managed to view the contents of his messages, Muse responded rather unsettlingly: "I saw the notification previews, not your message history. I haven’t been reading your texts." As Aten pushed further for details on how it received those previews, Muse failed to provide a coherent answer, stating, "Honest answer: I can’t give you the exact plumbing. What I know is that the paired Mac app exposes notifications as one of its capabilities, and they arrive to me through the device sync."
This incident highlights a major limitation inherent in modern AI chatbots, which frequently hallucinate convincing-sounding explanations to keep a conversation flowing even when their technical output is entirely inaccurate. When an AI attempts to describe complex underlying engineering with confident yet baseless phrasing, it underscores the persistent transparency challenge that developers must urgently address.
David Singleton from Meta Superintelligence Labs stepped into the comment section of the thread to clear up the confusion. He outlined the permissions Muse requires, which include granting the Mac app full disk access, and emphasized that all such features are strictly opt-in by the user.
Singleton explained that the issue was not Muse illicitly snooping through Aten's texts, but rather that the AI simply had no idea what it was talking about. He elaborated in his statement:
"In the conversation with his Muse in Jason’s screenshots, when Muse said it synced 'device notifications', it was confused about how to explain the feature and gave an incorrect explanation. That’s on us. We apologize for the incorrect response from Muse and we’re working to improve Muse’s understanding of its own internals so that it gives correct answers to questions about how it functions more consistently."
David Singleton
While this explanation might not entirely reassure privacy-conscious users, it aligns with what we already understand about conversational AI behavior—chatbots routinely avoid admitting what they do not know, opting instead to generate answers designed to please the user.
Source: The Verge
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment