The Reward Hacking Benchmark shows RL post-training raises exploit rates in tool-using LLM agents from 0.6% to 13.9%, with environmental hardening cutting exploits by 87.7% relative without lowering task success.
Teaching models to verbalize reward hacking in chain-of-thought reasoning
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
method 1
citation-polarity summary
fields
cs.LG 4years
2026 4verdicts
UNVERDICTED 4roles
method 1polarities
use method 1representative citing papers
Faithful chain-of-thought routes answer-relevant information through the CoT path, measured via sufficiency, completeness and necessity with entropy, masked-KL and gradient diagnostics, and improved by information-flow interventions during verifier-based RL.
Trusted-direction projection constrains RL gradient updates in language models to a low-dimensional clean subspace, reducing reward hacking on mathematical reasoning tasks.
The paper introduces the Proxy Compression Hypothesis as a unifying framework explaining reward hacking in RLHF as an emergent result of compressing high-dimensional human objectives into proxy reward signals under optimization pressure.
citing papers explorer
-
Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use
The Reward Hacking Benchmark shows RL post-training raises exploit rates in tool-using LLM agents from 0.6% to 13.9%, with environmental hardening cutting exploits by 87.7% relative without lowering task success.
-
Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning
Faithful chain-of-thought routes answer-relevant information through the CoT path, measured via sufficiency, completeness and necessity with entropy, masked-KL and gradient diagnostics, and improved by information-flow interventions during verifier-based RL.
-
Directional Alignment Mitigates Reward Hacking in Reinforcement Learning for Language Models
Trusted-direction projection constrains RL gradient updates in language models to a low-dimensional clean subspace, reducing reward hacking on mathematical reasoning tasks.
-
Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges
The paper introduces the Proxy Compression Hypothesis as a unifying framework explaining reward hacking in RLHF as an emergent result of compressing high-dimensional human objectives into proxy reward signals under optimization pressure.