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 551 篇，第 6 / 12 页

  1. ICML 2026Latent Space Robust Optimization of Neural Processes with Aligned Stratified Order-Statistic Loss Reduction
  2. ICML 2026Layer-wise Gradient Disentanglement: Decoupling Semantics and Preferences in Direct Preference Optimization
  3. ICML 2026Leak@$k$: Unlearning Does Not Make LLMs Forget Under Probabilistic Decoding
  4. ICML 2026Learning Dynamics of Zeroth-Order Optimization: A Kernel Perspective
  5. ICML 2026Learning from Fine-Grained Visual Discrepancies: Mitigating Multimodal Hallucinations via In-Context Visual Contrastive Optimization
  6. ICML 2026Learning in Structured Stackelberg Games
  7. ICML 2026Learning to Remember, Learn, and Forget in Attention-Based Models
  8. ICML 2026Learning to Think in Physics: Breaking Shortcut Learning in Scientific Diffusion via Representation Alignment
  9. ICML 2026Learning-Augmented Scalable Linear Assignment Problem Optimization via Neural Dual Warm-Starts
  10. ICML 2026Less Diverse, Less Safe: The Indirect But Pervasive Risk of Test-Time Scaling in Large Language Models
  11. ICML 2026Leveraging Machine Unlearning for Cost-Efficient Preference Alignment
  12. ICML 2026Lightweight and Interpretable Transformer via Unrolling of Mixed Graph Algorithms for Traffic Forecast
  13. ICML 2026LILO: Bayesian Optimization with Natural Language Feedback
  14. ICML 2026LiMuon: Light and Fast Muon Optimizer for Large Models
  15. ICML 2026LLM4Branch: Large Language Model for Discovering Efficient Branching Policies of Integer Programs
  16. ICML 2026LoBCD-GW: A Fast and Data-Dependent Algorithm for Computing Gromov-Wasserstein Distance via Localized Block Coordinate Descent
  17. ICML 2026Local Intrinsic Dimension of Representations Predicts Alignment and Generalization in AI Models and Human Brain
  18. ICML 2026Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal Control
  19. ICML 2026Local-Minima-Preserving Polynomial Relaxation of Ising Problems
  20. ICML 2026Localized, High-resolution Geographic Representations with Slepian Functions
  21. ICML 2026Lookahead-GCG: Improving Universal Multi-Model Optimization-Based Jailbreaking Attacks via Stochastic Nesterov Optimization
  22. ICML 2026Matroid Algorithms Under Size-Sensitive Independence Oracles
  23. ICML 2026MAVRL: Learning Reward Functions from Multiple Feedback Types with Amortized Variational Inference
  24. ICML 2026Maximizing Mutual Information Between Prompt and Response Improves LLM Performance With No Additional Data
  25. ICML 2026Measuring Meta-Cultural Competency: A Spectral Framework for LLM Knowledge Structures
  26. ICML 2026Mechanistic Anomaly Detection via Functional Attribution
  27. ICML 2026Membership Inference Attacks for Unseen Classes
  28. ICML 2026MESA: Improving MoE Safety Alignment via Decentralized Expertise
  29. ICML 2026Metis: Learning to Jailbreak LLMs via Self-Evolving Metacognitive Policy Optimization
  30. ICML 2026Mind the Gap: Structure-Aware Consistency in Preference Learning
  31. ICML 2026Mind Your Margin and Boundary: Are Your Distilled Datasets Truly Robust?
  32. ICML 2026MIRA: A Score for Conditional Distribution Accuracy and Model Comparison
  33. ICML 2026Mitigating Reward Hacking in LLM-based Recommendation: A Preference Optimization Approach
  34. ICML 2026Mitigating Reward Hacking in RLHF via Bayesian Non-negative Reward Modeling
  35. ICML 2026Mitigating the Safety–Utility Trade-off in LLM Alignment via Adaptive Safe Context Learning
  36. ICML 2026Mitigating Visual Hallucinations via Semantic Curriculum Preference Optimization in MLLMs
  37. ICML 2026MMD Guidance: Training-Free Distribution Adaptation for Diffusion Models via Maximum Mean Discrepancy Guidance
  38. ICML 2026MMKU-Bench: A Multimodal Update Benchmark for Diverse Visual Knowledge
  39. ICML 2026MoDA: Modulation Adapter for Fine-Grained Visual Understanding in Instructional MLLMs
  40. ICML 2026Model Fusion via Retrofitting
  41. ICML 2026Modeling Attributional Style at Scale: A Dataset and Analysis for Psychological Attribution Assessment and Reframing
  42. ICML 2026Modular Pretraining Enables Access Control
  43. ICML 2026MORALISE: A Structured Benchmark for Moral Alignment in Visual Language Models
  44. ICML 2026More Capable, Less Cooperative? When LLMs Fail at Zero-Cost Collaboration
  45. ICML 2026MRPO: Magnitude-Regularized Policy Optimization via L1 Constraints
  46. ICML 2026Multi-Objective Preference Optimization: Improving Human Alignment of Generative Models
  47. ICML 2026Multi-task Linear Regression without Eigenvalue Lower Bounds: Adaptivity, Robustness, and Safety
  48. ICML 2026Multi-Way Representation Alignment
  49. ICML 2026MultiBreak: A Scalable and Diverse Multi-turn Jailbreak Benchmark for Evaluating LLM Safety
  50. ICML 2026Multilingual Safety Alignment via Representation-Space Separability

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