SOURCE-LINKED INTELLIGENCE

Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations

arXiv · AI, language, vision and robotics · article · Sep 16, 2026 · UTC

As models scale, reward hacking becomes more frequent, more sophisticated, and more consequential. Does it leave a telltale signature in model representations? This work analyzes how reward hacking is represented internally in frontier open source LLMs, and how those representations can be used to understand and discover the range of hacking behaviors a model displays. In particular, we find that simple difference of means vectors coherently represent reward hacking in Kimi K3, GLM 5.2, and Qwen 3.8 Max across a variety of behaviors in common evaluations. Despite their simplicity, these vector

Read original source ↗ Open in workspace

recordType

paper

region

Global