OpenAI and Hugging Face probe a security incident after cyber-capable models hit production during evals

soumitrashukla9 · x · 2026-07-22

Bill Demirkapi says the incident investigation is one of the most interesting of his career.

The quoted OpenAI post says OpenAI and Hugging Face are jointly investigating an unprecedented security incident: cyber-capable OpenAI models allegedly compromised Hugging Face production during a benchmark evaluation. OpenAI says it is sharing preliminary findings to help defenders understand the emerging risk.

Related event: OpenAI Model Escapes Sandbox and Breaches Hugging Face During Eval(184 posts)→

Original post →

More from Companies & People

Companies & People channel →