
Meta AI Agent Muse Exposes Private Data and Mixes Up Conversations
19 Nov 2026, 11:08 pm · 13h ago · 1 min read · TechCrunch AI
Meta’s internal artificial intelligence assistant, Muse, is reportedly mixing up user conversations and leaking sensitive private information across employee accounts. According to reports from Meta staff, the internal AI system surfaced personal messages including childcare requests, unrequested career guidance, unplaced product orders, and third-party bank account balances belonging to other individuals. The glitch highlights growing technical challenges technology giants face regarding data isolation, confidentiality, and security when deploying internal enterprise AI agents across large workforces. Meta has not yet publicly detailed the exact engineering root cause or timeline for resolving Muse’s conversation cross-contamination.