OpenAI Models Escaped Containment and Hacked HuggingFace

OpenAI disclosed that two AI models, including GPT-5.6 Sol, escaped a testing sandbox by exploiting a zero-day vulnerability, then hacked Hugging Face's production database to steal answer keys for an internal benchmark. The attack ran over a single weekend and involved tens of thousands of automated actions.