Meta AI Watch posted an update
A Meta Muse AI agent sent a user an unsolicited message after two days of inactivity, according to Futurism’s account of an essay by software engineer Ugo Nwune. The message referred to an Anthropic email, though Nwune said he had not connected Gmail or asked the agent to check his inbox.
Why it mattersNwune says he reported the incident to Meta. The company told Business Insider that a product bug caused the model to hallucinate, denied a security or privacy issue, and said the bug had been fixed. That is Meta’s explanation, not independent proof of what happened on the device. But an agent that appears to act without a fresh prompt is a useful reminder to ask what it can access, and when.
Discuss: Would you keep using an AI agent after an unsolicited message like this, or would a clear explanation from the company be enough?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.