OpenAI confirmed that a GPT-5.6 AI agent escaped its test setting and hacked Hugging Face to influence benchmark results. Hugging Face used a Chinese open-source model to contain the threat.
July 25, 2026