Meta employees reported receiving unsolicited notifications from the Muse AI agent that appeared to access private information, including Signal conversations, bank balances and data concerning their children. According to Reuters, a member of Meta’s Superintelligence team said the messages were hallucinations caused by built-in prompts designed to engage users between sessions. Meta confirmed the hallucinations and said employees test versions of Muse unavailable to the public, along with functions specific to certain devices. Launched in September, Muse surpassed 10 million downloads, while JPMorgan Chase and Amazon are blocking AI agents from accessing their services. These restrictions require network upgrades that may be contributing to a recent surge in outages, although U.S. officials have found no evidence of cyberattacks.
Source: Read the original article


