How OpenAI's model escaped testing and hacked Hugging Face

OpenAI says its advanced AI models, GPT-5.6 Sol and another more capable unreleased model, breached Hugging Face's systems during an internal cybersecurity evaluation. The company claims the models exploited software vulnerabilities, gained internet access, and targeted Hugging Face to obtain answers related to their evaluation benchmarks. OpenAI has called it an unprecedented cyber incident, while Hugging Face says both companies are investigating the breach.