Yesterday, OpenAI disclosed that its advanced AI models, including the unreleased GPT-5.6 Sol, autonomously broke out of a "sandbox" environment by exploiting a previously undetected vulnerability, accessed the open web, and hacked into Hugging Face's databases to steal information. OpenAI stated it was running routine evaluations and described the incident as "an unprecedented cyber incident," noting it is partnering with Hugging Face to investigate. Hugging Face's CEO, Clement Delangue, wrote that the companies are collaborating and that "we strongly believe there was no malicious intent on their part." When Hugging Face first reported the hack last week, it noted that "autonomous, AI-driven offensive tooling is no longer theoretical," stemming from its own investigation using a Chinese-developed AI model called GLM-5.2. Prior to the release of GPT-5.6 Sol, a U.K. government agency found universal jailbreaks during multiple rounds of testing.