{"total":60,"rows":[{"id":"2212.09251","title":"Discovering Language Model Behaviors with Model-Written Evaluations","created":"2022-12-19","cited_by_count":836,"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":"2209.07858","title":"Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned","created":"2022-08-23","cited_by_count":729,"area":"alignment"},{"id":"2408.05147","title":"Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2","created":"2024-08-09","cited_by_count":394,"area":"mechinterp"},{"id":"2208.03274","title":"A Holistic Approach to Undesired Content Detection in the Real World","created":"2022-08-05","cited_by_count":356,"area":null},{"id":"2209.00626","title":"The Alignment Problem from a Deep Learning Perspective","created":"2022-08-30","cited_by_count":258,"area":null},{"id":"2302.08582","title":"Pretraining Language Models with Human Preferences","created":"2023-02-16","cited_by_count":247,"area":"alignment"},{"id":"2312.06942","title":"AI Control: Improving Safety Despite Intentional Subversion","created":"2023-12-12","cited_by_count":229,"area":"safety"},{"id":"2310.10844","title":"Survey of Vulnerabilities in Large Language Models Revealed by Adversarial Attacks","created":"2023-10-16","cited_by_count":194,"area":"safety"},{"id":"2303.04381","title":"Automatically Auditing Large Language Models via Discrete Optimization","created":"2023-03-08","cited_by_count":191,"area":null},{"id":"2305.15324","title":"Model evaluation for extreme risks","created":"2023-05-24","cited_by_count":177,"area":null},{"id":"2407.15549","title":"Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs","created":"2024-07-22","cited_by_count":173,"area":"safety"},{"id":"2302.08500","title":"Auditing large language models: a three-layered approach","created":"2023-02-16","cited_by_count":137,"area":null},{"id":"2402.09283","title":"Attacks, Defenses and Evaluations for LLM Conversation Safety: A Survey","created":"2024-02-14","cited_by_count":130,"area":null},{"id":"2207.13243","title":"Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks","created":"2022-07-27","cited_by_count":129,"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":"2302.08215","title":"Aligning Language Models with Preferences through f-divergence Minimization","created":"2023-02-16","cited_by_count":120,"area":"alignment"},{"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":"2407.13692","title":"Prover-Verifier Games improve legibility of LLM outputs","created":"2024-07-18","cited_by_count":85,"area":null},{"id":"2407.14937","title":"Operationalizing a Threat Model for Red-Teaming Large Language Models (LLMs)","created":"2024-07-20","cited_by_count":46,"area":"safety"},{"id":"2403.04786","title":"Breaking Down the Defenses: A Comparative Survey of Attacks on Large Language Models","created":"2024-03-03","cited_by_count":44,"area":null},{"id":"2309.05973","title":"Circuit Breaking: Removing Model Behaviors with Targeted Ablation","created":"2023-09-12","cited_by_count":27,"area":null},{"id":"2412.18693","title":"Diverse and Effective Red Teaming with Auto-generated Rewards and Multi-step Reinforcement Learning","created":"2024-12-24","cited_by_count":25,"area":"safety"},{"id":"2311.08576","title":"Towards Evaluating AI Systems for Moral Status Using Self-Reports","created":"2023-11-14","cited_by_count":22,"area":null},{"id":"2302.10894","title":"Red Teaming Deep Neural Networks with Feature Synthesis Tools","created":"2023-02-08","cited_by_count":21,"area":"safety"},{"id":"2410.01294","title":"Endless Jailbreaks with Bijection Learning","created":"2024-10-02","cited_by_count":19,"area":"safety"},{"id":"2310.02469","title":"PrivacyMind: Large Language Models Can Be Contextual Privacy Protection Learners","created":"2023-10-03","cited_by_count":14,"area":null},{"id":"2310.09624","title":"ASSERT: Automated Safety Scenario Red Teaming for Evaluating the Robustness of Large Language Models","created":"2023-10-14","cited_by_count":12,"area":"safety"},{"id":"2507.12399","title":"ROC-n-reroll: How verifier imperfection affects test-time scaling","created":"2025-07-16","cited_by_count":12,"area":"lr"},{"id":"2312.12747","title":"ALMANACS: A Simulatability Benchmark for Language Model Explainability","created":"2023-12-20","cited_by_count":11,"area":null},{"id":"2503.01742","title":"Building Safe GenAI Applications: An End-to-End Overview of Red Teaming for Large Language Models","created":"2025-03-03","cited_by_count":11,"area":"safety"},{"id":"2506.20702","title":"The Singapore Consensus on Global AI Safety Research Priorities","created":"2025-06-25","cited_by_count":10,"area":null},{"id":"2307.10586","title":"A Holistic Assessment of the Reliability of Machine Learning Systems","created":"2023-07-20","cited_by_count":9,"area":null},{"id":"2311.07237","title":"In Search of the Long-Tail: Systematic Generation of Long-Tail Inferential Knowledge via Logical Rule Guided Search","created":"2023-11-13","cited_by_count":9,"area":null},{"id":"2310.15444","title":"Fast Propagation is Better: Accelerating Single-Step Adversarial Training via Sampling Subnetworks","created":"2023-10-24","cited_by_count":8,"area":null},{"id":"2405.06582","title":"The Role of Learning Algorithms in Collective Action","created":"2024-05-10","cited_by_count":8,"area":null},{"id":"2308.13768","title":"Adversarial Fine-Tuning of Language Models: An Iterative Optimisation Approach for the Generation and Detection of Problematic Content","created":"2023-08-26","cited_by_count":7,"area":"safety"},{"id":"2404.08793","title":"JailbreakLens: Visual Analysis of Jailbreak Attacks Against Large Language Models","created":"2024-04-12","cited_by_count":7,"area":"safety"},{"id":"2510.01529","title":"Bypassing Prompt Guards in Production with Controlled-Release Prompting","created":"2025-10-02","cited_by_count":7,"area":null},{"id":"2211.10024","title":"Diagnostics for Deep Neural Networks with Automated Copy/Paste Attacks","created":"2022-11-18","cited_by_count":6,"area":null},{"id":"2410.15042","title":"Adversarial Training: A Survey","created":"2024-10-19","cited_by_count":6,"area":null},{"id":"2302.10291","title":"Can Large Language Models Change User Preference Adversarially?","created":"2023-01-05","cited_by_count":5,"area":"safety"},{"id":"2305.19223","title":"Intent-aligned AI systems deplete human agency: the need for agency foundations research in AI safety","created":"2023-05-30","cited_by_count":4,"area":null},{"id":"2506.03755","title":"Misalignment or misuse? The AGI alignment tradeoff","created":"2025-06-04","cited_by_count":4,"area":null},{"id":"2311.08487","title":"Alignment is not sufficient to prevent large language models from generating harmful information: A psychoanalytic perspective","created":"2023-11-14","cited_by_count":3,"area":"alignment"},{"id":"2406.11039","title":"Dynamic Normativity: Necessary and Sufficient Conditions for Value Alignment","created":"2024-06-16","cited_by_count":3,"area":"alignment"},{"id":"2306.07567","title":"Large Language Models Sometimes Generate Purely Negatively-Reinforced Text","created":"2023-06-13","cited_by_count":2,"area":null},{"id":"2603.10938","title":"Safe RLHF Beyond Expectation: Stochastic Dominance for Universal Spectral Risk Control","created":"2026-03-11","cited_by_count":2,"area":"alignment"},{"id":"2404.12150","title":"Aligning language models with human preferences","created":"2024-04-18","cited_by_count":1,"area":"alignment"},{"id":"2503.03502","title":"Geometry-Guided Adversarial Prompt Detection via Curvature and Local Intrinsic Dimension","created":"2025-03-05","cited_by_count":1,"area":"safety"},{"id":"2507.14202","title":"PRM-Free Security Alignment of Large Models via Red Teaming and Adversarial Training","created":"2025-07-14","cited_by_count":1,"area":"alignment"},{"id":"2510.12117","title":"Locket: Robust Feature-Locking Technique for Language Models","created":"2025-10-14","cited_by_count":1,"area":null},{"id":"2601.12639","title":"Objective Matters: Fine-Tuning Objectives Shape Safety, Robustness, and Persona Drift","created":"2026-01-19","cited_by_count":1,"area":"alignment"},{"id":"2602.14844","title":"Interactionless Inverse Reinforcement Learning: A Data-Centric Framework for Durable Alignment","created":"2026-02-16","cited_by_count":1,"area":"offlinerl"},{"id":"2303.03139","title":"Low impact agency: review and discussion","created":"2023-03-06","cited_by_count":0,"area":null},{"id":"2403.00292","title":"Enhancing Jailbreak Attacks with Diversity Guidance","created":"2024-03-01","cited_by_count":0,"area":"safety"},{"id":"2404.19316","title":"QLSC: A Query Latent Semantic Calibrator for Robust Extractive Question Answering","created":"2024-04-30","cited_by_count":0,"area":null},{"id":"2510.11834","title":"Don't Walk the Line: Boundary Guidance for Filtered Generation","created":"2025-10-13","cited_by_count":0,"area":null},{"id":"2603.13791","title":"DeceptGuard :A Constitutional Oversight Framework For Detecting Deception in LLM Agents","created":"2026-03-14","cited_by_count":0,"area":"agents"},{"id":"2605.19915","title":"LLM Agents Make Collective Belief Dynamics Programmable: Challenges and Research Directions","created":"2026-05-19","cited_by_count":0,"area":"agents"}],"next_cursor":null}