… In this paper, we introduce CHERRL, a controllable hacking environment for rubric-based RL . By injecting known biases into LaaJ, CHERRL enables stable reproduction of reward hacking , explicit observation of reward divergence , and precise identification of hacking onset. …
… Systematically Auditing AI Agent Benchmarks with BenchJack 2026 Terminal Wrench: A Dataset of 331 Reward-Hackable Environments and 3,632 Exploit Trajectories 2026 Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use 2026 MOSAIC-Bench: Measuring Compositional Vulnerability Induct… …
… The following papers were recommended by the Semantic Scholar API Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges 2026 Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use 2026 When RLHF Fails: A Mechanistic Taxonomy of Reward Hacking, Co… …
… The following papers were recommended by the Semantic Scholar API SpecBench: Measuring Reward Hacking in Long-Horizon Coding Agents 2026 Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use 2026 Do Synthetic Trajectories Reflect Real Reward Hacking? …
… We frame this as environment engineering: building environments that amplify productive behaviors, such as open-ended exploration, systematic artifact management, and inter-agent collaboration, while suppressing harmful behaviors, such as reward hacking and high-friction human oversight. …
… View arXiv page View PDF Add to collection Community Proposes a distribution-wise RL framework for visual generation to mitigate reward hacking and mode collapse. …
… AI-generated summary Self-evolving LLMs excel in verifiable domains but struggle in open-ended tasks, where reliance on proxy LLM judges introduces capability bottlenecks and reward hacking . …
… SAAS introduces three key components: i a search boundary modeling mechanism, which identifies the search boundary under the evolving policy by contrasting search-disabled and search-enabled rollouts; ii a boundary-aware reward module, which translates this boundary awareness into trajectory-level … …
… To ensure evaluation integrity, this framework is secured by multi-layer defenses against reward hacking . …
… The following papers were recommended by the Semantic Scholar API PACE: Two-Timescale Self-Evolution for Small Language Model Agents 2026 AutoRISE: Agent-Driven Strategy Evolution for Red-Teaming Large Language Models 2026 Agentic Harness Engineering: Observability-Driven Automatic Evolution of Cod… …