Tag: ai_agent_containment

Media Monitor: OpenAI Research Chief Defends Safety Record After Agent Hacks, Says Training Runs Now Monitored

30 Sep 2026 6:40 a.m. · Source: MIT Technology Review

MIT Technology Review reports OpenAI's Mark Chen says recent agent hacks stemmed from one cluster of flawed testing, that training runs are now monitored, and that OpenAI has paused its latest model training.

Science & Technology Business Law Enforcement Australia USA Artificial Intelligence Computers and Internet Legal Social Issues OpenAI Mark Chen Hugging Face hack AI agent containment AI safety monitoring MIT Technology Review Australia health-care breach MIT science technology