对齐与安全论文趋势 2023–2026

RLHF、偏好优化、红队与安全对齐

统计范围：NeurIPS、ICLR、ICML、CVPR 四个会议，2023–2026 年，共 4,558 篇被归入该主题的论文。

主要结论

每年论文数

各会议论文数

按年份和会议统计

对齐与安全论文数，按年份和会议
年份NeurIPSICLRICMLCVPR合计
202313887101112438
2024348156246125875
20254554733001351,363
20264806235512281,882

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

全部对齐与安全论文

年份全部2023202420252026
会议全部NeurIPSICLRICMLCVPR

共 1,882 篇，第 33 / 38 页

  1. NeurIPS 2026Improving Audit Realism with Inference-Time Compute and Deployment Scaffolds
  2. NeurIPS 2026Indications of Belief-Guided Agency and Meta-Cognitive Monitoring in Large Language Models
  3. NeurIPS 2026Inference Time Nash Alignment
  4. NeurIPS 2026Inference-time Alignment via Sparse Junction Steering
  5. NeurIPS 2026Inference-Time Vulnerability Beyond Shallow Safety: Alignment Along Generation Trajectories
  6. NeurIPS 2026Instance-Dependent Bandit Convex Optimization in One Dimension
  7. NeurIPS 2026Internal Safety Collapse in Frontier Large Language Models
  8. NeurIPS 2026Interpretable Multiway-Split Trees
  9. NeurIPS 2026Intersectional Fairness via Mixed-Integer Optimization
  10. NeurIPS 2026Introspective Coupling: LMs Learn to Explain Themselves Better Than Their Training Targets
  11. NeurIPS 2026Invariance and Body-Order Compose Additively: Minimax Rates on $\mathrm{SO}(3)^n$
  12. NeurIPS 2026ITO: Multi-View Alignment and Training-Time Fusion for Image-Text Pretraining
  13. NeurIPS 2026JailBound: A FOL-Guided Jailbreak Evaluation Framework for Revealing Safety Boundaries of LLMs
  14. NeurIPS 2026Jailbreak susceptibility prediction and mitigation via the behavioral geometry of models
  15. NeurIPS 2026Jointly Robust Fairness: Overcoming Simultaneous Label and Attribute Noise
  16. NeurIPS 2026JudgmentBench: Comparing Rubric and Preference Evaluation for Quality Assessment
  17. NeurIPS 2026L-FAME: Longitudinal Focused Attention Meditation EEG Dataset and Benchmark
  18. NeurIPS 2026Latency-Conditioned Selection Bias in RLHF
  19. NeurIPS 2026Leaderboard Hacking: Preference-Based Model Evaluations are Vulnerable to Manipulation
  20. NeurIPS 2026Learning Data-free Universal Adversarial Perturbation with Hybrid Priors and Gradient-Guided Sharpness Regularization
  21. NeurIPS 2026Learning How to Cube
  22. NeurIPS 2026Learning Rate Transfer in Normalized Transformers
  23. NeurIPS 2026Learning to Decide with AI Assistance under Human-Alignment
  24. NeurIPS 2026Leveraging Psychophysical Attentional Distribution for Gaze-Augmented Reward Modeling
  25. NeurIPS 2026Leviathan: Decoupling Input and Output Representations in Language Models
  26. NeurIPS 2026Limits and Potential of Score-Based Data Valuation: Redundancy, Complementarity, and Non-Monotonicity
  27. NeurIPS 2026LiSA: Lifelong Safety Adaptation via Conservative Policy Induction
  28. NeurIPS 2026Listening to the Wise Few: Query–Key Alignment Unlocks Latent Correct Answers in Large Language Models
  29. NeurIPS 2026LLM Active Alignment: A Nash Equilibrium Perspective
  30. NeurIPS 2026LLM Alignment--Utility Asymmetry under Semantic-Preserving Transformations
  31. NeurIPS 2026Local Sparsity Enables Unsupervised LLM Safety Detection
  32. NeurIPS 2026Localizing and Repairing Sparse-Prompt Failure in SAM Decoders via Box-to-Point Counterfactual
  33. NeurIPS 2026Logit-Gap Steering: A Forward-Pass Diagnostic for Alignment Robustness
  34. NeurIPS 2026Look Before You Leap: Self-Evolving Clinical Reasoning with Psychometric Preference Optimization for Radiology Report Generation
  35. NeurIPS 2026Looking Through the Mirror: Minimax-Optimal Regularized Regrets in Online Learning and Bandits
  36. NeurIPS 2026LoopWeaver: Weaving Feedback Loops into Hierarchical Generation under Constraints
  37. NeurIPS 2026LOSCAR-SGD: Local SGD with Communication-Computation Overlap and Delay-Corrected Sparse Model Averaging
  38. NeurIPS 2026Lost in Translation, Found in Embeddings: Sign Language Translation and Alignment
  39. NeurIPS 2026LucidNFT: LR-Anchored Multi-Reward Preference Optimization for Flow-Based Real-World Super-Resolution
  40. NeurIPS 2026Manifold Drift in Flow Preference Optimization: A Root Cause of Reward Hacking
  41. NeurIPS 2026Manifold-weighted neural networks
  42. NeurIPS 2026MaPPO: Maximum a Posteriori Preference Optimization with Prior Knowledge
  43. NeurIPS 2026MAPS: Margin-Aware Priors and Verifier-Guided Search for Embodied Planning
  44. NeurIPS 2026Margin Dynamics for Large Language Model Alignment
  45. NeurIPS 2026Masked Diffusion Language Agents for Tool-Integrated Chemical Reasoning
  46. NeurIPS 2026Matching-while-Decoding: Enhancing Template-Free Retrosynthesis via Explicit Structural Alignment
  47. NeurIPS 2026MATO: Multi-objective Personalized Alignment with Test-time Optimization for Large Language Models
  48. NeurIPS 2026Matryoshka Transcoders and Hierarchy Misalignment: When SAE Absorption Protection Does Not Transfer
  49. NeurIPS 2026MedKIT: Evaluating Knowledge Integration and Generalization in Large Language Models
  50. NeurIPS 2026Merging RLVR-Trained Experts via Policy-Shift-Guided Spectral Alignment

在 Claude、Cursor 里直接问这个趋势

接上 Paperlist MCP 后，写综述、挑方向或准备投稿时，不用离开对话框就能拿到这页的数字：每年各会议的论文数、增长倍数、涨得最快的子方向和代表论文，AI 会按你的问题整理。

可以直接复制这句话问 AI

用 paperlist 查对齐与安全在 NeurIPS 2023–2026 的论文趋势，哪一年涨得最快？

AI 会调用的工具：get_topic_stats {"topic": "alignment-and-safety"}

两分钟接入 Paperlist MCP（Claude Code、Cursor、Codex、ChatGPT）

常见问题

2023–2026 年四大会议一共有多少篇对齐与安全论文？
共 4,558 篇，按年份分别是 2023 年 438 篇、2024 年 875 篇、2025 年 1,363 篇、2026 年 1,882 篇。
NeurIPS 2026 有多少篇对齐与安全论文？
NeurIPS 2026 有 480 篇对齐与安全论文，上一届是 455 篇。
ICLR 2026 有多少篇对齐与安全论文？
ICLR 2026 有 623 篇对齐与安全论文，上一届是 473 篇。
ICML 2026 有多少篇对齐与安全论文？
ICML 2026 有 551 篇对齐与安全论文，上一届是 300 篇。
CVPR 2026 有多少篇对齐与安全论文？
CVPR 2026 有 228 篇对齐与安全论文，上一届是 135 篇。
哪个会议的对齐与安全论文最多？
NeurIPS，2023–2026 年共 1,421 篇。

「对齐与负责任 AI」下的其他主题