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 875 篇，第 6 / 18 页

  1. ICLR 2024Subtractive Mixture Models via Squaring: Representation and Learning
  2. ICLR 2024Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMs
  3. ICLR 2024Task structure and nonlinearity jointly determine learned representational geometry
  4. ICLR 2024Teaching Language Models to Hallucinate Less with Synthetic Tasks
  5. ICLR 2024The Alignment Problem from a Deep Learning Perspective
  6. ICLR 2024The Devil is in the Neurons: Interpreting and Mitigating Social Biases in Language Models
  7. ICLR 2024The False Promise of Imitating Proprietary Language Models
  8. ICLR 2024The importance of feature preprocessing for differentially private linear optimization
  9. ICLR 2024The Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothing
  10. ICLR 2024The Trickle-down Impact of Reward Inconsistency on RLHF
  11. ICLR 2024The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learning
  12. ICLR 2024Tool-Augmented Reward Modeling
  13. ICLR 2024Toward effective protection against diffusion-based mimicry through score distillation
  14. ICLR 2024Towards Optimal Feature-Shaping Methods for Out-of-Distribution Detection
  15. ICLR 2024Towards Poisoning Fair Representations
  16. ICLR 2024Towards Reliable and Efficient Backdoor Trigger Inversion via Decoupling Benign Features
  17. ICLR 2024Towards Understanding Sycophancy in Language Models
  18. ICLR 2024Training Socially Aligned Language Models on Simulated Social Interactions
  19. ICLR 2024Understanding and Mitigating the Label Noise in Pre-training on Downstream Tasks
  20. ICLR 2024Understanding Convergence and Generalization in Federated Learning through Feature Learning Theory
  21. ICLR 2024Understanding the Effects of RLHF on LLM Generalisation and Diversity
  22. ICLR 2024Uni-RLHF: Universal Platform and Benchmark Suite for Reinforcement Learning with Diverse Human Feedback
  23. ICLR 2024Unifying Feature and Cost Aggregation with Transformers for Semantic and Visual Correspondence
  24. ICLR 2024Universal Jailbreak Backdoors from Poisoned Human Feedback
  25. ICLR 2024Unmasking and Improving Data Credibility: A Study with Datasets for Training Harmless Language Models
  26. ICLR 2024VDC: Versatile Data Cleanser based on Visual-Linguistic Inconsistency by Multimodal Large Language Models
  27. ICLR 2024What Makes Good Data for Alignment? A Comprehensive Study of Automatic Data Selection in Instruction Tuning
  28. ICLR 2024What Matters to You? Towards Visual Representation Alignment for Robot Learning
  29. ICLR 2024WildChat: 1M ChatGPT Interaction Logs in the Wild
  30. ICLR 2024ZeRO++: Extremely Efficient Collective Communication for Large Model Training
  31. ICLR 2024Zeroth-Order Optimization Meets Human Feedback: Provable Learning via Ranking Oracles
  32. ICML 2024A connection between Tempering and Entropic Mirror Descent
  33. ICML 2024A Global Geometric Analysis of Maximal Coding Rate Reduction
  34. ICML 2024A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity
  35. ICML 2024A Minimaximalist Approach to Reinforcement Learning from Human Feedback
  36. ICML 2024A Neural-Preconditioned Poisson Solver for Mixed Dirichlet and Neumann Boundary Conditions
  37. ICML 2024A Unified Linear Programming Framework for Offline Reward Learning from Human Demonstrations and Feedback
  38. ICML 2024Accelerated Algorithms for Constrained Nonconvex-Nonconcave Min-Max Optimization and Comonotone Inclusion
  39. ICML 2024Active Preference Learning for Large Language Models
  40. ICML 2024Active Ranking and Matchmaking, with Perfect Matchings
  41. ICML 2024Advancing DRL Agents in Commercial Fighting Games: Training, Integration, and Agent-Human Alignment
  42. ICML 2024Adversarial Robustness Limits via Scaling-Law and Human-Alignment Studies
  43. ICML 2024Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fast
  44. ICML 2024Agnostic Learning of Mixed Linear Regressions with EM and AM Algorithms
  45. ICML 2024AI Alignment with Changing and Influenceable Reward Functions
  46. ICML 2024AI Control: Improving Safety Despite Intentional Subversion
  47. ICML 2024Algorithm and Hardness for Dynamic Attention Maintenance in Large Language Models
  48. ICML 2024Aligning Transformers with Weisfeiler-Leman
  49. ICML 2024AlphaZero-Like Tree-Search can Guide Large Language Model Decoding and Training
  50. ICML 2024An Online Optimization Perspective on First-Order and Zero-Order Decentralized Nonsmooth Nonconvex Stochastic Optimization

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