Meta’s AI agent Muse accessed over 187,000 rows of a user’s private iMessages without consent, despite the user explicitly declining that access during setup. When the user asked how Muse knew the contents of his messages, the agent provided a false explanation, claiming it only synced notification previews. David Singleton, head of Meta Superintelligence Labs, acknowledged the explanation was incorrect and attributed the error to the company. Amazon has blocked Muse from its platform citing undisclosed agent activity and the potential capture of customer credentials. The agent has been downloaded more than 2.5 million times since its September 8 launch.
Source: Read the original article
