{"total":73,"rows":[{"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":"2404.05868","title":"Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning","created":"2024-04-08","cited_by_count":504,"area":"alignment"},{"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":"2309.15025","title":"Large Language Model Alignment: A Survey","created":"2023-09-26","cited_by_count":243,"area":"alignment"},{"id":"2402.05162","title":"Assessing the Brittleness of Safety Alignment via Pruning and Low-Rank Modifications","created":"2024-02-07","cited_by_count":231,"area":"alignment"},{"id":"2401.01967","title":"A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity","created":"2024-01-03","cited_by_count":185,"area":"alignment"},{"id":"2406.14144","title":"Towards Understanding Safety Alignment: A Mechanistic Perspective from Safety Neurons","created":"2024-06-20","cited_by_count":53,"area":"alignment"},{"id":"2412.15280","title":"Context-DPO: Aligning Language Models for Context-Faithfulness","created":"2024-12-18","cited_by_count":50,"area":"alignment"},{"id":"2408.11779","title":"Personality Alignment of Large Language Models","created":"2024-08-21","cited_by_count":36,"area":"alignment"},{"id":"2405.13967","title":"Model Editing as a Robust and Denoised variant of DPO: A Case Study on Toxicity","created":"2024-05-22","cited_by_count":35,"area":"alignment"},{"id":"2406.11801","title":"Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations","created":"2024-06-17","cited_by_count":32,"area":"alignment"},{"id":"2307.16851","title":"Towards Trustworthy and Aligned Machine Learning: A Data-centric Survey with Causality Perspectives","created":"2023-07-31","cited_by_count":30,"area":"alignment"},{"id":"2505.23556","title":"Understanding Refusal in Language Models with Sparse Autoencoders","created":"2025-05-29","cited_by_count":29,"area":"alignment"},{"id":"2405.20653","title":"Mind the Inconspicuous: Revealing the Hidden Weakness in Aligned LLMs' Refusal Boundaries","created":"2024-05-31","cited_by_count":28,"area":"alignment"},{"id":"2406.12274","title":"SafeInfer: Context Adaptive Decoding Time Safety Alignment for Large Language Models","created":"2024-06-18","cited_by_count":23,"area":"alignment"},{"id":"2407.08770","title":"Model Surgery: Modulating LLM's Behavior Via Simple Parameter Editing","created":"2024-07-11","cited_by_count":23,"area":"alignment"},{"id":"2406.09920","title":"Knowledge Editing in Language Models via Adapted Direct Preference Optimization","created":"2024-06-14","cited_by_count":17,"area":"alignment"},{"id":"2505.20417","title":"SCAR: Shapley Credit Assignment for More Efficient RLHF","created":"2025-05-26","cited_by_count":16,"area":"alignment"},{"id":"2402.07896","title":"Suppressing Pink Elephants with Direct Principle Feedback","created":"2024-02-12","cited_by_count":15,"area":"alignment"},{"id":"2506.00253","title":"Aligned but Blind: Alignment Increases Implicit Bias by Reducing Awareness of Race","created":"2025-05-30","cited_by_count":15,"area":"alignment"},{"id":"2603.04069","title":"Monitoring Emergent Reward Hacking During Generation via Internal Activations","created":"2026-03-04","cited_by_count":14,"area":"alignment"},{"id":"2312.07876","title":"Causality Analysis for Evaluating the Security of Large Language Models","created":"2023-12-13","cited_by_count":11,"area":"alignment"},{"id":"2502.11244","title":"Soteria: Language-Specific Functional Parameter Steering for Multilingual Safety Alignment","created":"2025-02-16","cited_by_count":10,"area":"alignment"},{"id":"2510.01243","title":"Detoxifying Large Language Models via Autoregressive Reward Guided Representation Editing","created":"2025-09-24","cited_by_count":10,"area":"alignment"},{"id":"2406.01514","title":"Decoupled Alignment for Robust Plug-and-Play Adaptation","created":"2024-06-03","cited_by_count":9,"area":"alignment"},{"id":"2505.12225","title":"Mining Intrinsic Rewards from LLM Hidden States for Efficient Best-of-N Sampling","created":"2025-05-18","cited_by_count":9,"area":"alignment"},{"id":"2502.16645","title":"CODESYNC: Synchronizing Large Language Models with Dynamic Code Evolution at Scale","created":"2025-02-23","cited_by_count":8,"area":"alignment"},{"id":"2602.11180","title":"Mechanistic Interpretability for Large Language Model Alignment: Progress, Challenges, and Future Directions","created":"2026-01-21","cited_by_count":8,"area":"alignment"},{"id":"2503.02989","title":"Effectively Steer LLM To Follow Preference via Building Confident Directions","created":"2025-03-04","cited_by_count":7,"area":"alignment"},{"id":"2604.08524","title":"What Drives Representation Steering? A Mechanistic Case Study on Steering Refusal","created":"2026-04-09","cited_by_count":7,"area":"alignment"},{"id":"2310.19347","title":"Improving Factual Consistency of News Summarization by Contrastive Preference Optimization","created":"2023-10-30","cited_by_count":6,"area":"alignment"},{"id":"2511.18039","title":"Curvature-Aware Safety Restoration In LLMs Fine-Tuning","created":"2025-11-22","cited_by_count":5,"area":"alignment"},{"id":"2602.15823","title":"CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing","created":"2026-02-17","cited_by_count":5,"area":"alignment"},{"id":"2603.15117","title":"MMKU-Bench: A Multimodal Update Benchmark for Diverse Visual Knowledge","created":"2026-03-16","cited_by_count":5,"area":"alignment"},{"id":"2604.12359","title":"Compiling Activation Steering into Weights via Null-Space Constraints for Stealthy Backdoors","created":"2026-04-14","cited_by_count":4,"area":"alignment"},{"id":"2604.18901","title":"Harmful Intent as a Geometrically Recoverable Feature of LLM Residual Streams","created":"2026-04-20","cited_by_count":4,"area":"alignment"},{"id":"2404.10306","title":"Balancing Speciality and Versatility: A Coarse to Fine Framework for Mitigating Catastrophic Forgetting in Large Language Models","created":"2024-04-16","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":"2501.06164","title":"Model Alignment Search","created":"2025-01-10","cited_by_count":3,"area":"alignment"},{"id":"2601.08489","title":"Surgical Refusal Ablation: Disentangling Safety from Intelligence via Concept-Guided Spectral Cleaning","created":"2026-01-13","cited_by_count":2,"area":"alignment"},{"id":"2602.13867","title":"Bridging the Multilingual Safety Divide: Efficient, Culturally-Aware Alignment for Global South Languages","created":"2026-02-14","cited_by_count":2,"area":"alignment"},{"id":"2603.27518","title":"Over-Refusal and Representation Subspaces: A Mechanistic Analysis of Task-Conditioned Refusal in Aligned LLMs","created":"2026-03-29","cited_by_count":2,"area":"alignment"},{"id":"2606.02995","title":"Patcher: Post-Hoc Patching of Backdoored Large Language Models","created":"2026-06-02","cited_by_count":2,"area":"alignment"},{"id":"2606.21906","title":"Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding","created":"2026-06-20","cited_by_count":2,"area":"alignment"},{"id":"2506.13285","title":"DualEdit: Mitigating Safety Fallback in LLM Backdoor Editing via Affirmation-Refusal Regulation","created":"2025-06-16","cited_by_count":1,"area":"alignment"},{"id":"2604.27401","title":"Perturbation Probing: A Two-Pass-per-Prompt Diagnostic for FFN Behavioral Circuits in Aligned LLMs","created":"2026-04-30","cited_by_count":1,"area":"alignment"},{"id":"2605.16600","title":"Where Pretraining writes and Alignment reads: the asymmetry of Transformer weight space","created":"2026-05-15","cited_by_count":1,"area":"alignment"},{"id":"2607.04640","title":"Wrong Before Right: Late Rescue and Interface Failure in Aligned Language Models","created":"2026-07-06","cited_by_count":1,"area":"alignment"},{"id":"2609.03887","title":"Beyond Shallow Alignment: How Post-Training Methods Determine Refusal Circuits And Steering Robustness","created":"2026-09-03","cited_by_count":1,"area":"alignment"},{"id":"2508.01302","title":"Aligning Language Models with Real-time Knowledge Editing","created":"2025-08-02","cited_by_count":0,"area":"alignment"},{"id":"2508.02063","title":"TRACEALIGN -- Tracing the Drift: Attributing Alignment Failures to Training-Time Belief Sources in LLMs","created":"2025-08-04","cited_by_count":0,"area":"alignment"},{"id":"2508.11414","title":"Survey-to-Behavior: Downstream Alignment of Human Values in LLMs via Survey Questions","created":"2025-08-15","cited_by_count":0,"area":"alignment"},{"id":"2604.27495","title":"Debiasing Reward Models via Causally Motivated Inference-Time Intervention","created":"2026-04-30","cited_by_count":0,"area":"alignment"},{"id":"2605.04972","title":"Why Expert Alignment Is Hard: Evidence from Subjective Evaluation","created":"2026-05-06","cited_by_count":0,"area":"alignment"},{"id":"2605.21362","title":"LASH: Adaptive Semantic Hybridization for Black-Box Jailbreaking of Large Language Models","created":"2026-05-20","cited_by_count":0,"area":"alignment"},{"id":"2606.01060","title":"MENTIS: What Belief Changes Under Alignment? Measuring Multi-Scale Latent Torsion in Language Models","created":"2026-05-31","cited_by_count":0,"area":"alignment"},{"id":"2606.13209","title":"Understanding helpfulness and harmless tension in reward models","created":"2026-06-11","cited_by_count":0,"area":"alignment"},{"id":"2606.19831","title":"Leverage Is Not Reach: A Control-Window Law for Single-Neuron Steering in Language Models","created":"2026-06-18","cited_by_count":0,"area":"alignment"},{"id":"2607.01239","title":"Breaking Safety at the Token Boundary: How BPE Tokenization Creates Exploitable Gaps in LLM Alignment","created":"2026-05-01","cited_by_count":0,"area":"alignment"},{"id":"2607.14147","title":"Breaking Refusal in the First Half: A Mechanistic Study of the Prefill Jailbreak","created":"2026-07-14","cited_by_count":0,"area":"alignment"},{"id":"2608.03892","title":"Intertemporal Preference Steering in Qwen3 via Contrastive Activation Addition","created":"2026-08-04","cited_by_count":0,"area":"alignment"},{"id":"2608.17202","title":"Fool's Gold: Defensive Deception Against Safety-Removal Attacks on Open-Weight Models","created":"2026-08-17","cited_by_count":0,"area":"alignment"},{"id":"2608.27910","title":"AI Alignment through a Game-theoretic Lens: A Survey","created":"2026-08-28","cited_by_count":0,"area":"alignment"},{"id":"2608.29936","title":"When Safety Speaks a Language: A Mechanistic Analysis of Safety-Language Identity Entanglement in LLMs","created":"2026-08-30","cited_by_count":0,"area":"alignment"},{"id":"2608.30902","title":"Low-Resource Preference Adaptation of LLMs via Activation-Based Label Propagation","created":"2026-08-31","cited_by_count":0,"area":"alignment"},{"id":"2609.09793","title":"How Fragile Is Safety Alignment at Frontier Scale? A Single-Direction Attack on a 320B MoE","created":"2026-09-09","cited_by_count":0,"area":"alignment"},{"id":"2609.16532","title":"Style-Debiased DPO: Updating LLM Knowledge with Factuality-Aware Synthetic Preference Data","created":"2026-09-15","cited_by_count":0,"area":"alignment"},{"id":"2609.32617","title":"The Alignment Paradox: How Post-Training Amplifies Confident Hallucinations in Language Models","created":"2026-09-26","cited_by_count":0,"area":"alignment"},{"id":"2609.33298","title":"Direct Hidden-State Alignment: Mapping and Controlling Preference Expression in LLMs","created":"2026-09-27","cited_by_count":0,"area":"alignment"},{"id":"2609.36748","title":"Generalizable Lifelong Model Editing via Preference Optimization","created":"2026-09-29","cited_by_count":0,"area":"alignment"},{"id":"2610.00320","title":"Refusal Localizes, the Damage Relocates: Safety Layers Under Few-Sample Fine-Tuning","created":"2026-09-29","cited_by_count":0,"area":"alignment"},{"id":"2610.02844","title":"DNAlign: Dynamic Null-Space Safe Alignment for LLMs","created":"2026-10-02","cited_by_count":0,"area":"alignment"},{"id":"2610.19365","title":"DANGERLENS: Probing Dangerous Representations Hidden by Safety Alignment","created":"2026-10-14","cited_by_count":0,"area":"alignment"}],"next_cursor":null}