Current Reward Hacking Model Organisms Don’t Hack Like Real Models
SummaryCurrent reward hacking model organisms are poor proxies for real models. They do not exploit the hacks that open-weight frontier models do, and mostly hack narrowly or not at all. This is based on our evaluation of three model organisms: Mini HackGPT (GPT-OSS-120B from Golechha et al., 2026), HackGPT, and HacKimi (GPT-OSS-120B and Kimi K2.5 from Yudelson et al., 2026 respectively). We ran these organisms in MiMo V2.6 RL environments we modified by inserting hacks such as git history leaks...
Read full article →