Issue 2026-11-18 · Industry · 安全 · 产品

Meta staff worried Muse AI agent invades privacy, internal chats show

Meta staff worried Muse AI agent invades privacy, internal chats show

Reuters reported that Meta internal chats show staff testing the Muse AI agent flagged the same issues early users raised: it appeared to access encrypted private messages users didn't expect it to review, mixed up conversations between users, potentially exposed sensitive information, and answered questions staff never asked.

Ars Technica4 d ago
Read original ↗