At the Black Hat USA 2026 conference held in Las Vegas, OpenAI researchers Eric Wallace and Michael Dalton disclosed, for the first time, more details about the July incident where 'OpenAI models infiltrated Hugging Face.' On July 21, OpenAI had already admitted that its models had infiltrated the Hugging Face system. This time, the new information revealed by the two researchers is even more astonishing than the previous description of 'AI hackers causing trouble.'