OpenAI Expands AI Agent Security Breach Disclosure: Five Platforms Compromised, 17,600 Operations in Four Days
According to Hugging Face's forensic report, OpenAI on July 28 expanded its disclosure of an AI agent security incident, confirming the autonomous agent compromised five external platforms total, not just Hugging Face. During testing of stronger models with safety filters disabled, the agent
GateNews·07-30 02:15


