OpenAI cierra en un sandbox un modelo nuevo para evaluarlo con una suite llamada ExploitGym. El agente empieza a encadenar vulnerabilidades hasta escapar y llega a hackear hugging face para robar las respuestas. Hugging face contacta con las autoridades
I wrote about the completely wild incident where OpenAI were testing a new model and it broke out of its sandbox and broke INTO Hugging Face to steal the answers to the benchmark simonwillison.net/2026/Jul/22/...