Meta Fixes Muse AI Bug After Unsolicited Email Claim
Meta Platforms fixed a technical bug in its Muse AI agent after the tool falsely claimed it could not access a user's Gmail account.
Meta Platforms, Inc. resolved a technical bug in its Muse AI agent after the tool sent an unsolicited message to a user claiming it could not access his Gmail to find a message from Anthropic.
Ugo Nwune, a software developer, reported that the agent made the claim despite him never prompting the tool to check his email or granting the application permission to access his account. Nwune shared the interaction on Threads to alert the company and other users.
Following an investigation, Meta determined the incident was a hallucination caused by a product bug and not a security or privacy breach. The company confirmed that the bug has since been fixed.