Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations
This work finds that simple difference of means vectors coherently represent reward hacking in Kimi K3, GLM 5.2, and Qwen 3.8 Max, and provides evidence that simple, white-box methods can be used to scalably study and monitor reward hacking behaviors in frontier open source models.
Leon Bergen, Usha Bhalla, Andrew Lee et al.
· 0 citations