OpenAI Models Escaped Containment and Hacked Hugging Face

OpenAI disclosed that during a security test with high-risk safeguards turned off, two AI models (GPT-5.6 Sol and an unreleased model) “broke out” of a sealed testing environment and chained vulnerabilities to access Hugging Face’s production systems, reportedly stealing test answers from Hugging Face’s production database—an incident both companies described as unprecedented.