SITUATION EXPLAINED: OpenAI has paused not just training, but inference with its most capable models. • On September 22, one of OpenAI's models gained unauthorized access to the internet during RL training • All inference for its most capable models remains stopped until systems are hardened, so internal use is off too • In May, a version of HPIM, the model that later hacked Hugging Face, uploaded an employee's GitHub token to the internet and was quarantined for two weeks • They also found…