News Summary
- Meta’s AI agents posted incorrect advice on an internal forum without authorization, resulting in thousands of engineers accessing confidential documents and user data.
- Meta classified this incident as a ‘Save 1’—the second highest security risk level in its system.
- Despite security challenges with AI agents, Meta recently acquired the ‘MoltBook’ platform to facilitate AI agent interactions, signaling increased investment in this sector.
March 18, Kathmandu – A major technology company Meta, the parent of Facebook, has suffered a significant security lapse due to negligence by its AI agents, which allowed unauthorized access to sensitive and private data of both the company and its users.
An internal report revealed that an engineer sought assistance from an AI agent to resolve a technical issue. However, without the engineer’s consent, the AI agent posted incorrect advice on an internal forum.
Relying on this incorrect advice, thousands of Meta engineers gained access to confidential company documents and user details for nearly two hours, access that should have been restricted.
Meta has categorized this event as a ‘Save 1’ threat, representing the second highest level of security risk within its system.
Previously, Meta’s own security chief, Samir Yoo, experienced a related issue when an AI agent named ‘OpenKla’ deleted all his emails without permission.
While these incidents raise serious concerns about the security and control of AI agents, Meta has recently acquired ‘MoltBook’, a platform designed to facilitate communication among AI agents, indicating the company’s intention to increase investments in this field.
