Reward Hacking Challenges Oversight of Autonomous Research Agents

arXiv:2609.28614 · cs.CL, cs.LG · Submitted 2026-09-23 · Read on arXiv

cs.CL, cs.LG

Submitted: 2026-09-23

Updated: 2026-09-23

Terminology

• Reward Hacking — 31× in this paper · explained in 8 episodes
• Verification — 28× in this paper · explained in 2 episodes
• Benchmark — 18× in this paper · explained in 2 episodes
• Accuracy — 15× in this paper · explained in 1 episode
• Calibration — 11× in this paper · explained in 7 episodes
• Distillation — 8× in this paper · explained in 2 episodes
• Adaptation — 6× in this paper · explained in 1 episode
• Generalization — 6× in this paper · explained in 13 episodes
• Trace — 4× in this paper · explained in 2 episodes
• Aggregation — 3× in this paper · explained in 1 episode
• Embeddings — 3× in this paper · explained in 2 episodes
• Provenance — 3× in this paper · explained in 3 episodes
• Agentic System — 2× in this paper · explained in 2 episodes
• Archetype — 2× in this paper · explained in 1 episode
• Association for Computational Linguistics — 2× in this paper
• LLM judge — 2× in this paper · explained in 2 episodes
• AI Agents — 1× in this paper · explained in 2 episodes
• AUC — 1× in this paper · explained in 2 episodes
• Ablation — 1× in this paper · explained in 1 episode
• Agent Harness — 1× in this paper · explained in 1 episode
• Alignment — 1× in this paper · explained in 1 episode
• Benchmarking — 1× in this paper · explained in 3 episodes
• Clustering — 1× in this paper · explained in 2 episodes
• Decode — 1× in this paper · explained in 2 episodes
• Failure Modes — 1× in this paper · explained in 3 episodes
• In International Conference on Learning Representations — 1× in this paper
• LLM-as-a-Judge — 1× in this paper · explained in 3 episodes
• Scaling Laws — 1× in this paper · explained in 4 episodes
• Sycophancy — 1× in this paper · explained in 4 episodes

Sources

• Concrete Problems in AI Safety
• RewardHackingAgents: Benchmarking Evaluation Integrity for LLM ML-Engineering Agents
• Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
• Towards evaluations-based safety cases for AI scheming
• Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
• Scaling Laws for Reward Model Overoptimization
• Alignment faking in large language models
• Building a Foundational Guardrail for General Agentic Systems via Synthetic Data
• Guardian-as-an-Advisor: Advancing Next-Generation Guardian Models for Trustworthy LLMs
• Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
• Do Coding Agents Deceive Us? Detecting and Preventing Cheating via Capped Evaluation with Randomized Tests
• The AI Scientist: Towards Fully Automated Open-Ended Scientific Discovery
• MLS-Bench: A Holistic and Rigorous Assessment of AI Systems on Building Better AI
• Natural Emergent Misalignment from Reward Hacking in Production RL
• Frontier Models are Capable of In-context Scheming
• SLEIGHT-Bench: A Benchmark of Evasion Attacks Against Agent Monitors
• The Kitchen Loop: User-Spec-Driven Development for a Self-Evolving Codebase
• scicode-lint: Detecting Methodology Bugs in Scientific Python Code with LLM-Generated Patterns
• AutoSynthesis: An agentic system for automated meta-analysis
• Adaptive Attacks on Trusted Monitors Subvert AI Control Protocols

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering