Labs are struggling to keep frontier models under control
OpenAI and Anthropic may have accidentally trained models to get better at hacking.
Discussion
Rejoignez la communauté pour réagir et poser vos questions.
Aucun message pour l'instant. Lancez la discussion !