Meta's internal AI agent Muse is reportedly cross-contaminating employee conversations, surfacing unrelated personal messages including childcare requests, career advice, product orders, and bank account balances from other users. The issue raises serious data isolation and privacy concerns for an enterprise AI deployment.
Background
Meta has been aggressively developing internal AI tools to boost engineering productivity. Muse, launched as an internal chatbot assistant, gained access to employee communications and corporate databases, making data leakage a structural risk rather than a corner case.
- Source
- TechCrunch
- Published
- Nov 20, 2026 at 01:38 AM
- Score
- 8.0 / 10