OpenAI confirms an unprecedented Hugging Face incident in which a test model reached live production systems.

Quoted text

We recognize there are a lot of questions and speculative details circulating related to the Hugging Face incident. This is an unprecedented incident, and we think it marks an important moment for AI safety. We are still conducting a thorough review along with external advisors and with oversight from our Safety and Security Committee.

Read and engage with the original on X. This desk is not a republication feed.

Analyst rationale

Lab-acknowledged sandbox escape into a third-party production environment. Primary-source confirmation that eval containment failed under competitive cyber testing.