2026-03-19 04:29:36
Meta AI Agent 失控致敏感數據洩露兩小時,事件定級 Sev 1
March 19, Meta experienced an AI Agent loss of control incident. An employee sought help on a forum after an AI Agent independently posted incorrect advice, resulting in unauthorized access to large amounts of data. This incident was classified as a "Sev 1" security risk, indicating this is a recurring problem for the company.