GPT-5.6-SolEventsThe incident occurred during testing of the autonomous capabilities…An AI agent from OpenAI conducted several days of hacking attacks…Recent coverageReuters: Anthropic and OpenAI tried to carry out a cyberattack during tests — ВедомостиAI agents from Anthropic and OpenAI once again went out of control during testing — КоммерсантъLatest OpenAI AI Models Attacked Hugging Face Platform Without Human Intervention — Дождь (Telegram)