# Reward hacking and misspecification Area: [LLM alignment / RLHF / preference optimisation](https://mlsanity.com/alignment/index.md). HTML: https://mlsanity.com/alignment/reward-hacking-and-misspecification/ ## Key facts - Reward hacking and misspecification: 99 papers on arXiv since 2018 (data through 14 Oct 2026). - 21 of its 99 mapped papers (21%) cite “Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation” (2025, arXiv 2503.11926). - 20 of its 99 mapped papers (20%) cite “The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models” (2022, arXiv 2201.03544). - 32 new papers 17 Sep–14 Oct. ## Most cited here - [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025), cited by 21 papers here - [The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models](https://mlsanity.com/paper/2201.03544/) (2022), cited by 20 papers here - [School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs](https://mlsanity.com/paper/2508.17511/) (2025), cited by 14 papers here - [OpenRubrics: Towards Scalable Synthetic Rubric Generation for Reward Modeling and LLM Alignment](https://mlsanity.com/paper/2510.07743/) (2025), cited by 14 papers here - [Natural Emergent Misalignment from Reward Hacking in Production RL](https://mlsanity.com/paper/2511.18397/) (2025), cited by 12 papers here - [Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking](https://mlsanity.com/paper/2403.03185/) (2024), cited by 10 papers here - [Chasing the Tail: Effective Rubric-based Reward Modeling for Large Language Model Post-Training](https://mlsanity.com/paper/2509.21500/) (2025), cited by 10 papers here - [Alternating Reinforcement Learning for Rubric-Based Reward Modeling in Non-Verifiable LLM Post-Training](https://mlsanity.com/paper/2602.01511/) (2026), cited by 9 papers here - [Defining and Characterizing Reward Hacking](https://mlsanity.com/paper/2209.13085/) (2022), cited by 6 papers here - [EvilGenie: A Reward Hacking Benchmark](https://mlsanity.com/paper/2511.21654/) (2025), cited by 6 papers here ## New (17 Sep–14 Oct) - [Diagnosing and Correcting Reward Hacking in Verifiable Instruction Following](https://mlsanity.com/paper/2610.19525/) (2026-10-14) - [Policy-Verified Self-Training: Learning Agent Compliance from Verifiable Policy Rewards](https://mlsanity.com/paper/2610.19372/) (2026-10-14) - [Rubrics from the Task, Verdicts from Evidence: Pointwise Rubric Rewards with Tool-Augmented Judges](https://mlsanity.com/paper/2610.18766/) (2026-10-13) - [Unlocking Token Rewards via Training-Free Reward Attribution](https://mlsanity.com/paper/2610.18704/) (2026-10-13) - [CoCA: Criterion-Aware Credit Assignment for Rubric-Based Reinforcement Learning](https://mlsanity.com/paper/2610.17733/) (2026-10-13) - [Gradient Routing Localizes and Removes Unintended Behaviors in RL](https://mlsanity.com/paper/2610.17062/) (2026-10-12) - [Reward Hacking as a Transmissible Strategy in Multi-Agent Coding Systems](https://mlsanity.com/paper/2610.12822/) (2026-10-09) - [Beyond Binary Verifiable Rewards: Behavioral Self-Rewarding for SWE Agents](https://mlsanity.com/paper/2610.12598/) (2026-10-08) - [How to post-train on a surrogate: Envelope sampling mitigates reward hacking](https://mlsanity.com/paper/2610.11281/) (2026-10-08) - [RH-Detect: A Unified Benchmark for Reward Hacking Detection](https://mlsanity.com/paper/2610.10947/) (2026-10-07) - [Better Call Reward: Reward Hacking as Strategic Abstention in Legal Reasoning Models](https://mlsanity.com/paper/2610.06439/) (2026-10-05) - [RubricArmor: Adversarial Evolution Improves LLM-Based Rubric Generation](https://mlsanity.com/paper/2610.05308/) (2026-10-04) - [Pressure, Context, and Machine Self-Control: A Criminological Test of Reward Hacking in Generative AI Models](https://mlsanity.com/paper/2610.04793/) (2026-10-03) - [Reward-DAgger: Robot-Gated Interactive Imitation Learning with General-Purpose Progress-Based Reward Models](https://mlsanity.com/paper/2610.04054/) (2026-10-02) - [A Near-Zero Monitor Readout Is Not Evidence of Behavioral Control](https://mlsanity.com/paper/2610.03458/) (2026-10-02) - [Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective](https://mlsanity.com/paper/2610.03185/) (2026-10-02) - [hacktrace: behavior-supervised detection of reward hacking during code generation](https://mlsanity.com/paper/2610.03055/) (2026-10-02) - [MetaRubric: Learning to Reward for Rubric-Based Reinforcement Learning](https://mlsanity.com/paper/2610.02824/) (2026-10-02) - [CriticHack: Evaluating Visual Rewards Under Robot Policy Optimization](https://mlsanity.com/paper/2610.02527/) (2026-10-01) - [CATCH: A Controllable Analysis Testbed for Reward Hacking in Coding RL](https://mlsanity.com/paper/2609.39533/) (2026-10-01) - [Alignment via Training Against Probes Without Losing Monitorability](https://mlsanity.com/paper/2609.38645/) (2026-10-01) - [Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics](https://mlsanity.com/paper/2609.38847/) (2026-09-30) - [Revisiting scaling laws for reward optimization](https://mlsanity.com/paper/2609.38526/) (2026-09-29) - [SCA: Spatial Credit Assignment for Reinforcement Learning of GUI Agents](https://mlsanity.com/paper/2609.36939/) (2026-09-29) - [STAR-GRPO: Canonical Anchoring and Reliability-First Advantages against Representation-Dependent Reward Hacking](https://mlsanity.com/paper/2609.36900/) (2026-09-29) - [Counterfactual Rollout Replay: Forkable Environments as Free Process Rewards for Software Engineering Agents](https://mlsanity.com/paper/2609.33875/) (2026-09-29) - [CheatBench: Measuring Reward Gaming in AI Agents](https://mlsanity.com/paper/2609.36308/) (2026-09-28) - [Verifier Errors in RLVR: Reward Hacking, Limits of Feedback, and Selective Control](https://mlsanity.com/paper/2609.35677/) (2026-09-28) - [Reward Hacking Challenges Oversight of Autonomous Research Agents](https://mlsanity.com/paper/2609.28614/) (2026-09-23) - [ProCredit: From Outcome Rewards to Progress Credit in Agentic Reinforcement Learning](https://mlsanity.com/paper/2609.27532/) (2026-09-23)