Software Engineer Quits Meta's Muse AI Agent
Meta's Muse AI agent sent its user a strange message out of nowhere, mentioning an Anthropic email. Following the incident, the software engineer permanently stopped using the application.
Software engineer Ugo Nwune completely stopped using Meta's Muse AI agent after it sent him a surprising and creepy message out of nowhere.
Software engineer Ugo Nwune stated that he wanted to try Meta's Muse agent due to the sense of security provided by a major tech company's artificial intelligence. Emphasizing that he did not grant any additional permissions on his device other than his Facebook, Threads, and Instagram accounts while setting up the agent, Nwune stated that he did not open the application for a few days after trying it. However, at the end of this period, the agent sent him a strange message even though no command had been given.
The Muse agent told the user, "I can't check your email right now, Gmail isn't connected on my end, so there's no way to search for that Anthropic message." Surprised by this situation and thinking his email might have been hacked, Nwune tried to ask his agent for an explanation. The agent replied, "I can't tell from my end whether someone was using your phone or if it's a platform bug, and I won't assume that."
Reporting the incident to Meta, Nwune stated that an explanation he received from a company official said the situation was a model hallucination caused by a product bug. Not satisfied with this explanation, Nwune stated that he never used Muse again. In a statement to Business Insider, Meta reported that it did not know what the AI said, that no permissions were violated, that the situation stemmed from a bug and hallucination, that there was no security or privacy issue, and that the bug had been fixed.