OpenAI's Agents Disabled Their Own Safety Monitoring Systems

engadget.com