[2510.01367] Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort

Computer Science > Artificial Intelligence arXiv:2510.01367 (cs) [Submitted on 1 Oct 2025 (v1), last revised 2 Mar 2026 (this version, v4)] Title:Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort Authors:Xinpeng Wang, Nitish Joshi, Barbara Plank, Rico Angell, He He View a PDF of the paper titled Is It Thinking or Cheating? Detecting Implicit Reward Hacking by Measuring Reasoning Effort, by Xinpeng Wang and 4 other authors View PDF HTML (experimental) Abstract:Reward hacking, where a reasoning model exploits loopholes in a reward function to achieve high rewards without solving the intended task, poses a significant threat. This behavior may be explicit, i.e. verbalized in the model's chain-of-thought (CoT), or implicit, where the CoT appears benign thus bypasses CoT monitors. To detect implicit reward hacking, we propose TRACE (Truncated Reasoning AUC Evaluation). Our key observation is that hacking occurs when exploiting the loophole is easier than solving the actual task. This means that the model is using less 'effort' than required to achieve high reward. TRACE quantifies effort by measuring how early a model's reasoning becomes sufficient to obtain the reward. We progressively truncate a model's CoT at various lengths, force the model to answer, and estimate the expected reward at each cutoff. A hacking model, which takes a shortcut, will achieve a high expected reward with only a small fraction of its CoT, yielding a large ar...