# Information-Theoretic Reward Modeling for Stable RLHF: Detecting and Mitigating Reward Hacking

arXiv: https://arxiv.org/abs/2510.13694 · HTML: https://mlsanity.com/paper/2510.13694/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Reward modeling and RLHF, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 71 of its 92 references that are in the map.
- First posted on arXiv 15 Oct 2025.

## Placed in

- LLM alignment / RLHF / preference optimisation / [Reward modeling and RLHF](https://mlsanity.com/alignment/reward-modeling-and-rlhf/)

## Builds on (most cited first, 30 of 71)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [TruthfulQA: Measuring How Models Mimic Human Falsehoods](https://mlsanity.com/paper/2109.07958/) (2021)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [WebGPT: Browser-assisted question-answering with human feedback](https://mlsanity.com/paper/2112.09332/) (2021)
- [Dream to Control: Learning Behaviors by Latent Imagination](https://mlsanity.com/paper/1912.01603/) (2019)
- [A Simple Unified Framework for Detecting Out-of-Distribution Samples and Adversarial Attacks](https://mlsanity.com/paper/1807.03888/) (2018)
- [ZeRO: Memory Optimizations Toward Training Trillion Parameter Models](https://mlsanity.com/paper/1910.02054/) (2019)
- [Scaling Laws for Reward Model Overoptimization](https://mlsanity.com/paper/2210.10760/) (2022)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [Deep Variational Information Bottleneck](https://mlsanity.com/paper/1612.00410/) (2016)
- [A General Theoretical Paradigm to Understand Learning from Human Preferences](https://mlsanity.com/paper/2310.12036/) (2023)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [Deep Learning and the Information Bottleneck Principle](https://mlsanity.com/paper/1503.02406/) (2015)
- [Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2307.15217/) (2023)
- [AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback](https://mlsanity.com/paper/2305.14387/) (2023)
- [The Flan Collection: Designing Data and Methods for Effective Instruction Tuning](https://mlsanity.com/paper/2301.13688/) (2023)
- [OpenAssistant Conversations -- Democratizing Large Language Model Alignment](https://mlsanity.com/paper/2304.07327/) (2023)
- [Baichuan 2: Open Large-scale Language Models](https://mlsanity.com/paper/2309.10305/) (2023)

## Built on by (most cited first, 5 of 5)

- [Towards Robust LLM Post-Training: Automatic Failure Management for Reinforcement Fine-Tuning](https://mlsanity.com/paper/2605.04431/) (2026)
- [Towards Robust Endogenous Reasoning: Unifying Drift Adaptation in Non-Stationary Tuning](https://mlsanity.com/paper/2604.15705/) (2026)
- [Factored Causal Representation Learning for Robust Reward Modeling in RLHF](https://mlsanity.com/paper/2601.21350/) (2026)
- [QuantumQA: Enhancing Scientific Reasoning via Physics-Consistent Dataset and Verification-Aware Reinforcement Learning](https://mlsanity.com/paper/2604.18176/) (2026)
- [Beyond One-Size-Fits-All: Diagnosis-Driven Online Reinforcement Learning with Offline Priors](https://mlsanity.com/paper/2606.25527/) (2026)

## Abstract (arXiv)

Despite the success of Reinforcement Learning from Human Feedback (RLHF) in aligning language models with human values, reward hacking-or reward over-optimization-remains a major challenge. We identify two key obstacles to its mitigation: (1) reward misgeneralization in reward modeling, where reward models overfit to spurious, preference-irrelevant features; and (2) the lack of suitable regularization during RL optimization, as existing token-level constraints often over-restrict the policy space. To address these issues, we propose InfoRM, an information-theoretic reward modeling framework based on the Information Bottleneck (IB) principle, which filters out preference-irrelevant information to alleviate reward misgeneralization. We further observe that reward-hacked responses manifest as pronounced outliers in InfoRM's IB latent space, measured by Mahalanobis distance from the SFT-induced distribution. Motivated by this, we introduce IBL, a distribution-level regularization that penalizes such deviations, effectively expanding the optimization landscape while maintaining alignment. We prove that IBL is theoretically equivalent to the pessimistic RL objective within the IB latent space. Finally, we present Mahalanobis Outlier Probability (MOP), a statistical metric for quantifying reward hacking severity, enabling principled hyperparameter tuning and online mitigation such as early stopping. Extensive experiments across diverse LLMs and datasets confirm the generality of our findings, the effectiveness of InfoRM and IBL, and the reliability of MOP as a diagnostic tool-collectively advancing the state of RLHF.