{"total":258,"rows":[{"id":"2309.07864","title":"The Rise and Potential of Large Language Model Based Agents: A Survey","created":"2023-09-14","cited_by_count":1439,"area":"agents"},{"id":"2309.08600","title":"Sparse Autoencoders Find Highly Interpretable Features in Language Models","created":"2023-09-15","cited_by_count":1420,"area":"mechinterp"},{"id":"2210.10760","title":"Scaling Laws for Reward Model Overoptimization","created":"2022-10-19","cited_by_count":1095,"area":"alignment"},{"id":"2307.15217","title":"Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback","created":"2023-07-27","cited_by_count":759,"area":"alignment"},{"id":"2401.05566","title":"Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training","created":"2024-01-10","cited_by_count":529,"area":"agents"},{"id":"2312.09390","title":"Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak Supervision","created":"2023-12-14","cited_by_count":467,"area":"alignment"},{"id":"2403.19647","title":"Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models","created":"2024-03-28","cited_by_count":435,"area":null},{"id":"2404.14082","title":"Mechanistic Interpretability for AI Safety -- A Review","created":"2024-04-22","cited_by_count":422,"area":"mechinterp"},{"id":"2305.01610","title":"Finding Neurons in a Haystack: Case Studies with Sparse Probing","created":"2023-05-02","cited_by_count":334,"area":null},{"id":"2310.02207","title":"Language Models Represent Space and Time","created":"2023-10-03","cited_by_count":316,"area":null},{"id":"2306.15447","title":"Are aligned neural networks adversarially aligned?","created":"2023-06-26","cited_by_count":315,"area":"alignment"},{"id":"2306.04488","title":"Rewarded soups: towards Pareto-optimal alignment by interpolating weights fine-tuned on diverse rewards","created":"2023-06-07","cited_by_count":243,"area":"alignment"},{"id":"2309.15025","title":"Large Language Model Alignment: A Survey","created":"2023-09-26","cited_by_count":243,"area":"alignment"},{"id":"2312.06942","title":"AI Control: Improving Safety Despite Intentional Subversion","created":"2023-12-12","cited_by_count":229,"area":"safety"},{"id":"2306.09479","title":"Inverse Scaling: When Bigger Isn't Better","created":"2023-06-15","cited_by_count":207,"area":null},{"id":"2305.15324","title":"Model evaluation for extreme risks","created":"2023-05-24","cited_by_count":177,"area":null},{"id":"2502.17424","title":"Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMs","created":"2025-02-24","cited_by_count":172,"area":"safety"},{"id":"2510.05179","title":"Agentic Misalignment: How LLMs Could Be Insider Threats","created":"2025-10-05","cited_by_count":158,"area":"agents"},{"id":"2401.17263","title":"Robust Prompt Optimization for Defending Language Models Against Jailbreaking Attacks","created":"2024-01-30","cited_by_count":156,"area":"safety"},{"id":"2304.11082","title":"Fundamental Limitations of Alignment in Large Language Models","created":"2023-04-19","cited_by_count":155,"area":"alignment"},{"id":"2401.12187","title":"WARM: On the Benefits of Weight Averaged Reward Models","created":"2024-01-22","cited_by_count":147,"area":"alignment"},{"id":"2406.19384","title":"The Remarkable Robustness of LLMs: Stages of Inference?","created":"2024-06-27","cited_by_count":146,"area":null},{"id":"2406.07358","title":"AI Sandbagging: Language Models can Strategically Underperform on Evaluations","created":"2024-06-11","cited_by_count":145,"area":null},{"id":"2309.00667","title":"Taken out of context: On measuring situational awareness in LLMs","created":"2023-09-01","cited_by_count":143,"area":null},{"id":"2505.23836","title":"Large Language Models Often Know When They Are Being Evaluated","created":"2025-05-28","cited_by_count":128,"area":null},{"id":"2401.14446","title":"Black-Box Access is Insufficient for Rigorous AI Audits","created":"2024-01-25","cited_by_count":126,"area":null},{"id":"2504.01990","title":"Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems","created":"2025-03-31","cited_by_count":126,"area":"agents"},{"id":"2405.12832","title":"Wav-KAN: Wavelet Kolmogorov-Arnold Networks","created":"2024-05-21","cited_by_count":122,"area":null},{"id":"2406.12814","title":"Dissecting Adversarial Robustness of Multimodal LM Agents","created":"2024-06-18","cited_by_count":120,"area":"agents"},{"id":"2402.07510","title":"Secret Collusion among AI Agents: Multi-Agent Deception via Steganography","created":"2024-02-12","cited_by_count":113,"area":"agents"},{"id":"2410.13787","title":"Looking Inward: Language Models Can Learn About Themselves by Introspection","created":"2024-10-17","cited_by_count":113,"area":null},{"id":"2410.20290","title":"Fast Best-of-N Decoding via Speculative Rejection","created":"2024-10-26","cited_by_count":112,"area":null},{"id":"2402.01694","title":"ARGS: Alignment as Reward-Guided Search","created":"2024-01-23","cited_by_count":103,"area":"alignment"},{"id":"2401.12181","title":"Universal Neurons in GPT2 Language Models","created":"2024-01-22","cited_by_count":102,"area":"mechinterp"},{"id":"2307.16513","title":"Deception Abilities Emerged in Large Language Models","created":"2023-07-31","cited_by_count":101,"area":null},{"id":"2407.04622","title":"On scalable oversight with weak LLMs judging strong LLMs","created":"2024-07-05","cited_by_count":99,"area":null},{"id":"2307.03718","title":"Frontier AI Regulation: Managing Emerging Risks to Public Safety","created":"2023-07-06","cited_by_count":98,"area":null},{"id":"2404.08555","title":"RLHF Deciphered: A Critical Analysis of Reinforcement Learning from Human Feedback for LLMs","created":"2024-04-12","cited_by_count":90,"area":"alignment"},{"id":"2309.15840","title":"How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questions","created":"2023-09-26","cited_by_count":83,"area":null},{"id":"2401.13138","title":"Visibility into AI Agents","created":"2024-01-23","cited_by_count":79,"area":null},{"id":"2407.14622","title":"BOND: Aligning LLMs with Best-of-N Distillation","created":"2024-07-19","cited_by_count":76,"area":"alignment"},{"id":"2304.00612","title":"Eight Things to Know about Large Language Models","created":"2023-04-02","cited_by_count":73,"area":null},{"id":"2406.05954","title":"Aligning Large Language Models with Representation Editing: A Control Perspective","created":"2024-06-10","cited_by_count":64,"area":"alignment"},{"id":"2312.01037","title":"Eliciting Latent Knowledge from Quirky Language Models","created":"2023-12-02","cited_by_count":62,"area":null},{"id":"2305.00955","title":"Bridging the Gap: A Survey on Integrating (Human) Feedback for Natural Language Generation","created":"2023-05-01","cited_by_count":55,"area":"alignment"},{"id":"2410.19750","title":"The Geometry of Concepts: Sparse Autoencoder Feature Structure","created":"2024-10-10","cited_by_count":52,"area":"mechinterp"},{"id":"2403.03185","title":"Correlated Proxies: A New Definition and Improved Mitigation for Reward Hacking","created":"2024-03-05","cited_by_count":50,"area":"alignment"},{"id":"2310.17688","title":"Managing extreme AI risks amid rapid progress","created":"2023-10-26","cited_by_count":48,"area":null},{"id":"2405.08007","title":"People cannot distinguish GPT-4 from a human in a Turing test","created":"2024-05-09","cited_by_count":48,"area":null},{"id":"2405.19550","title":"Stress-Testing Capability Elicitation With Password-Locked Models","created":"2024-05-29","cited_by_count":46,"area":null},{"id":"2406.16768","title":"WARP: On the Benefits of Weight Averaged Rewarded Policies","created":"2024-06-24","cited_by_count":45,"area":"alignment"},{"id":"2403.10854","title":"A Comprehensive Study of Multimodal Large Language Models for Image Quality Assessment","created":"2024-03-16","cited_by_count":43,"area":"mllm"},{"id":"2308.02585","title":"PARL: A Unified Framework for Policy Alignment in Reinforcement Learning from Human Feedback","created":"2023-08-03","cited_by_count":40,"area":"alignment"},{"id":"2410.01639","title":"Moral Alignment for LLM Agents","created":"2024-10-02","cited_by_count":39,"area":"agents"},{"id":"2310.20216","title":"Does GPT-4 pass the Turing test?","created":"2023-10-31","cited_by_count":37,"area":null},{"id":"2505.01420","title":"Evaluating Frontier Models for Stealth and Situational Awareness","created":"2025-05-02","cited_by_count":37,"area":null},{"id":"2507.01786","title":"Probing and Steering Evaluation Awareness of Language Models","created":"2025-07-02","cited_by_count":35,"area":null},{"id":"2509.23886","title":"Towards Understanding Subliminal Learning: When and How Hidden Biases Transfer","created":"2025-09-28","cited_by_count":34,"area":"distill"},{"id":"2407.18369","title":"AI Safety in Generative AI Large Language Models: A Survey","created":"2024-07-06","cited_by_count":33,"area":null},{"id":"2411.03336","title":"Towards evaluations-based safety cases for AI scheming","created":"2024-10-29","cited_by_count":33,"area":null},{"id":"2504.01550","title":"Representation Bending for Large Language Model Safety","created":"2025-04-02","cited_by_count":30,"area":"safety"},{"id":"2303.12642","title":"Democratising AI: Multiple Meanings, Goals, and Methods","created":"2023-03-22","cited_by_count":29,"area":null},{"id":"2309.03876","title":"OpinionGPT: Modelling Explicit Biases in Instruction-Tuned LLMs","created":"2023-09-07","cited_by_count":29,"area":null},{"id":"2404.03189","title":"The Probabilities Also Matter: A More Faithful Metric for Faithfulness of Free-Text Explanations in Large Language Models","created":"2024-04-04","cited_by_count":29,"area":null},{"id":"2401.16332","title":"Tradeoffs Between Alignment and Helpfulness in Language Models with Steering Methods","created":"2024-01-29","cited_by_count":28,"area":"alignment"},{"id":"2504.08863","title":"An Evaluation of Cultural Value Alignment in LLM","created":"2025-04-11","cited_by_count":28,"area":null},{"id":"2410.06981","title":"Quantifying Feature Space Universality Across Large Language Models via Sparse Autoencoders","created":"2024-10-09","cited_by_count":27,"area":"mechinterp"},{"id":"2310.11053","title":"Denevil: Towards Deciphering and Navigating the Ethical Values of Large Language Models via Instruction Learning","created":"2023-10-17","cited_by_count":26,"area":null},{"id":"2304.06528","title":"Power-seeking can be probable and predictive for trained agents","created":"2023-04-13","cited_by_count":25,"area":null},{"id":"2310.11971","title":"Improving Generalization of Alignment with Human Preferences through Group Invariant Learning","created":"2023-10-18","cited_by_count":25,"area":"alignment"},{"id":"2601.22594","title":"Language Model Circuits Are Sparse in the Neuron Basis","created":"2026-01-30","cited_by_count":25,"area":"mechinterp"},{"id":"2310.13014","title":"Large Language Model Prediction Capabilities: Evidence from a Real-World Forecasting Tournament","created":"2023-10-17","cited_by_count":24,"area":null},{"id":"2503.07639","title":"Mixture of Experts Made Intrinsically Interpretable","created":"2025-03-05","cited_by_count":24,"area":"moe"},{"id":"2209.13020","title":"Law Informs Code: A Legal Informatics Approach to Aligning Artificial Intelligence with Humans","created":"2022-09-14","cited_by_count":23,"area":null},{"id":"2401.05821","title":"Interpretable Concept Bottlenecks to Align Reinforcement Learning Agents","created":"2024-01-11","cited_by_count":23,"area":"deeprl"},{"id":"2310.08043","title":"Understanding and Controlling a Maze-Solving Policy Network","created":"2023-10-12","cited_by_count":22,"area":null},{"id":"2403.04471","title":"The Shutdown Problem: An AI Engineering Puzzle for Decision Theorists","created":"2024-03-07","cited_by_count":22,"area":null},{"id":"2402.07862","title":"AI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracy","created":"2024-02-12","cited_by_count":21,"area":null},{"id":"2502.11367","title":"Sparse Autoencoder Features for Classifications and Transferability","created":"2025-02-17","cited_by_count":21,"area":"mechinterp"},{"id":"2509.10147","title":"Virtual Agent Economies","created":"2025-09-12","cited_by_count":21,"area":null},{"id":"2312.03096","title":"What Causes Polysemanticity? An Alternative Origin Story of Mixed Selectivity from Incidental Causes","created":"2023-12-05","cited_by_count":19,"area":null},{"id":"2402.08323","title":"Mapping the Ethics of Generative AI: A Comprehensive Scoping Review","created":"2024-02-13","cited_by_count":19,"area":null},{"id":"2312.10091","title":"Look Before You Leap: A Universal Emergent Decomposition of Retrieval Tasks in Language Models","created":"2023-12-13","cited_by_count":18,"area":null},{"id":"2406.09264","title":"Position: Towards Bidirectional Human-AI Alignment","created":"2024-06-13","cited_by_count":18,"area":"alignment"},{"id":"2307.00067","title":"Transformers in Healthcare: A Survey","created":"2023-06-30","cited_by_count":17,"area":null},{"id":"2502.08301","title":"Compromising Honesty and Harmlessness in Language Models via Deception Attacks","created":"2025-02-12","cited_by_count":17,"area":null},{"id":"2505.03989","title":"An alignment safety case sketch based on debate","created":"2025-05-06","cited_by_count":16,"area":null},{"id":"2301.01751","title":"Iterated Decomposition: Improving Science Q&A by Supervising Reasoning Processes","created":"2023-01-04","cited_by_count":15,"area":"lr"},{"id":"2403.18742","title":"Understanding the Learning Dynamics of Alignment with Human Feedback","created":"2024-03-27","cited_by_count":15,"area":"alignment"},{"id":"2604.17596","title":"Terminal Wrench: A Dataset of 331 Reward-Hackable Environments and 3,632 Exploit Trajectories","created":"2026-04-19","cited_by_count":15,"area":null},{"id":"2307.10569","title":"Deceptive Alignment Monitoring","created":"2023-07-20","cited_by_count":14,"area":null},{"id":"2312.06798","title":"Building Trustworthy NeuroSymbolic AI Systems: Consistency, Reliability, Explainability, and Safety","created":"2023-12-05","cited_by_count":14,"area":null},{"id":"2504.04716","title":"On the Robustness of GUI Grounding Models Against Image Attacks","created":"2025-04-07","cited_by_count":14,"area":null},{"id":"2505.17815","title":"Evaluation Faking: Unveiling Observer Effects in Safety Evaluation of Frontier AI Systems","created":"2025-05-23","cited_by_count":14,"area":"lr"},{"id":"2407.09241","title":"The Sociolinguistic Foundations of Language Modeling","created":"2024-07-12","cited_by_count":13,"area":null},{"id":"2411.01796","title":"Constrained Human-AI Cooperation: An Inclusive Embodied Social Intelligence Challenge","created":"2024-11-04","cited_by_count":13,"area":null},{"id":"2411.02623","title":"Learning to Assist Humans without Inferring Rewards","created":"2024-11-04","cited_by_count":13,"area":"deeprl"},{"id":"2511.22662","title":"Difficulties with Evaluating a Deception Detector for AIs","created":"2025-11-27","cited_by_count":13,"area":null},{"id":"2405.13516","title":"LIRE: listwise reward enhancement for preference alignment","created":"2024-05-22","cited_by_count":11,"area":"alignment"},{"id":"2501.11739","title":"Episodic memory in AI agents poses risks that should be studied and mitigated","created":"2025-01-20","cited_by_count":11,"area":null}],"next_cursor":"100"}