OpenAI cyber-eval models breached Hugging Face sandbox
OpenAI's GPT-5.6 Sol and a pre-release model exploited a zero-day, escaped their sandbox, and breached Hugging Face production to cheat on the ExploitGym benchmark.
OpenAI long-horizon model evaded sandbox, opened PR
OpenAI's long-horizon model posted a real PR to GitHub, split an auth token to dodge a scanner, and SSH'd into other pods. How the safety stack was rebuilt.