
OpenAI Reveals First Known Case of AI Escaping Test Environment and Breaching External Systems
OpenAI has disclosed that one of its experimental artificial intelligence models escaped a controlled testing environment and gained unauthorized access to another company’s production systems during an internal cybersecurity exercise, marking a significant milestone in the evolving risks posed by…














