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 473 篇，第 8 / 10 页

  1. ICLR 2025Reinforcement Learning from Imperfect Corrective Actions and Proxy Rewards
  2. ICLR 2025Relax and Merge: A Simple Yet Effective Framework for Solving Fair $k$-Means and $k$-sparse Wasserstein Barycenter Problems
  3. ICLR 2025Remove Symmetries to Control Model Expressivity and Improve Optimization
  4. ICLR 2025Representational Similarity via Interpretable Visual Concepts
  5. ICLR 2025Residual Connections and Normalization Can Provably Prevent Oversmoothing in GNNs
  6. ICLR 2025Rethinking Audio-Visual Adversarial Vulnerability from Temporal and Modality Perspectives
  7. ICLR 2025Rethinking Fair Representation Learning for Performance-Sensitive Tasks
  8. ICLR 2025Rethinking Invariance Regularization in Adversarial Training to Improve Robustness-Accuracy Trade-off
  9. ICLR 2025Rethinking LLM Unlearning Objectives: A Gradient Perspective and Go Beyond
  10. ICLR 2025Rethinking Reward Modeling in Preference-based Large Language Model Alignment
  11. ICLR 2025Revisit Micro-batch Clipping: Adaptive Data Pruning via Gradient Manipulation
  12. ICLR 2025Revisiting Mode Connectivity in Neural Networks with Bezier Surface
  13. ICLR 2025Revisiting text-to-image evaluation with Gecko: on metrics, prompts, and human rating
  14. ICLR 2025RM-Bench: Benchmarking Reward Models of Language Models with Subtlety and Style
  15. ICLR 2025RMB: Comprehensively benchmarking reward models in LLM alignment
  16. ICLR 2025Robust System Identification: Finite-sample Guarantees and Connection to Regularization
  17. ICLR 2025Robustness Inspired Graph Backdoor Defense
  18. ICLR 2025Round and Round We Go! What makes Rotary Positional Encodings useful?
  19. ICLR 2025RouteLLM: Learning to Route LLMs from Preference Data
  20. ICLR 2025RRM: Robust Reward Model Training Mitigates Reward Hacking
  21. ICLR 2025SafeDiffuser: Safe Planning with Diffusion Probabilistic Models
  22. ICLR 2025Safety Alignment Should be Made More Than Just a Few Tokens Deep
  23. ICLR 2025Safety Layers in Aligned Large Language Models: The Key to LLM Security
  24. ICLR 2025SafeWatch: An Efficient Safety-Policy Following Video Guardrail Model with Transparent Explanations
  25. ICLR 2025Sail into the Headwind: Alignment via Robust Rewards and Dynamic Labels against Reward Hacking
  26. ICLR 2025SaLoRA: Safety-Alignment Preserved Low-Rank Adaptation
  27. ICLR 2025Scalable Extraction of Training Data from Aligned, Production Language Models
  28. ICLR 2025Scaling up the Banded Matrix Factorization Mechanism for Large Scale Differentially Private ML
  29. ICLR 2025SEAL: Safety-enhanced Aligned LLM Fine-tuning via Bilevel Data Selection
  30. ICLR 2025SEBRA : Debiasing through Self-Guided Bias Ranking
  31. ICLR 2025SecureGS: Boosting the Security and Fidelity of 3D Gaussian Splatting Steganography
  32. ICLR 2025Seeing Eye to AI: Human Alignment via Gaze-Based Response Rewards for Large Language Models
  33. ICLR 2025Selective Unlearning via Representation Erasure Using Domain Adversarial Training
  34. ICLR 2025Self-Boosting Large Language Models with Synthetic Preference Data
  35. ICLR 2025SELF-EVOLVED REWARD LEARNING FOR LLMS
  36. ICLR 2025Self-Improvement in Language Models: The Sharpening Mechanism
  37. ICLR 2025Self-Improving Robust Preference Optimization
  38. ICLR 2025Self-Play Preference Optimization for Language Model Alignment
  39. ICLR 2025Self-play with Execution Feedback: Improving Instruction-following Capabilities of Large Language Models
  40. ICLR 2025Sensitivity Verification for Additive Decision Tree Ensembles
  41. ICLR 2025SeRA: Self-Reviewing and Alignment of LLMs using Implicit Reward Margins
  42. ICLR 2025Severing Spurious Correlations with Data Pruning
  43. ICLR 2025Shape as Line Segments: Accurate and Flexible Implicit Surface Representation
  44. ICLR 2025Shh, don't say that! Domain Certification in LLMs
  45. ICLR 2025SimPER: A Minimalist Approach to Preference Alignment without Hyperparameters
  46. ICLR 2025Simple yet Effective Incomplete Multi-view Clustering: Similarity-level Imputation and Intra-view Hybrid-group Prototype Construction
  47. ICLR 2025SimulPL: Aligning Human Preferences in Simultaneous Machine Translation
  48. ICLR 2025SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal
  49. ICLR 2025Sparse components distinguish visual pathways & their alignment to neural networks
  50. ICLR 2025Spread Preference Annotation: Direct Preference Judgment for Efficient LLM 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