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 36 / 38 页

  1. NeurIPS 2026RL-Inf: Tracking Non-local Training Data Influence for Online Reinforcement Learning
  2. NeurIPS 2026Robust and Efficient Continual Model Merging via Global Singular Subspace Separation and Restoration
  3. NeurIPS 2026Robust and Scalable Collaborative Learning via Pull-Based Epidemic Communication
  4. NeurIPS 2026Robust Domain Generalization under Divergent Marginal and Conditional Distributions
  5. NeurIPS 2026Robust Policy Optimization to Prevent Catastrophic Forgetting
  6. NeurIPS 2026Root Cause Analysis of Measurement and Mechanistic Anomalies
  7. NeurIPS 2026Routing as a Singular Reparameterization: A Closed-Form Pushforward Prior and Exact Bayesian Complexity in a Minimal Proxy
  8. NeurIPS 2026RSSA: Robust Semantic and Spatial Aligner for Collaborative Perception
  9. NeurIPS 2026Rubric-Align: Safety Alignment through Dynamically Co-Evolving Rubrics
  10. NeurIPS 2026RxDiff: Discrete Diffusion for Medication Recommendation with Inference-Time Safety Control
  11. NeurIPS 2026S2T-RLHF: Hierarchical Credit Assignment for Stable Preference-Based RLHF
  12. NeurIPS 2026Safe in Its Own Words: Self-Guided Safety Alignment for Multimodal Reasoning Models
  13. NeurIPS 2026Safe-PG-LQR: Provably Safe, Convergent and Optimal Model-Free Linear Quadratic Regulator with Hard Constraints
  14. NeurIPS 2026Safe, or Simply Incapable? Rethinking Safety Evaluation for Phone-Use Agents
  15. NeurIPS 2026SafeDrug: A Benchmark Dataset for Safety-Critical Pharmacological Reasoning in LLMs
  16. NeurIPS 2026SafePro: Generation of Safe and Functional Proteins via Cross-Task Conflict-Aware Alignment
  17. NeurIPS 2026Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction
  18. NeurIPS 2026Safety Reconstructed: Generative Modeling via Masked Diffusion Builds Strong Safety Guardrails
  19. NeurIPS 2026SAMoR: Motion Modelling for Articulated Objects of Any Skeleton and Topology
  20. NeurIPS 2026SAMPPO: Structure-Aware Mirror Proximal Policy Optimization
  21. NeurIPS 2026SAVeR$^2$: Reasoning-based Safety Alignment for Large Reasoning Models via Verifiable Rewards
  22. NeurIPS 2026Scalable Distributed Stochastic Optimization via Bidirectional Compression: Beyond Pessimistic Limits
  23. NeurIPS 2026Scalable Neural Safety Certification via Monotonicity
  24. NeurIPS 2026Scaling Linear Mode Connectivity and Merging to Billion Parameter Pretrained Transformers
  25. NeurIPS 2026Screening Lipid Nanoparticles through Structure-Ratio Alignment
  26. NeurIPS 2026SeaPilot: Mobile Agent with Self-refining Environment Alignment
  27. NeurIPS 2026Seeing Through the Chain: Understanding and Mitigating Hallucinations in Multimodal Large Reasoning Models
  28. NeurIPS 2026Selective Disk Bispectrum: A Complete and Rotation Invariant Image Descriptor
  29. NeurIPS 2026Selective Safety Steering via Value-Filtered Decoding
  30. NeurIPS 2026Self-Recognition Finetuning can Reverse and Prevent Emergent Misalignment
  31. NeurIPS 2026SGD Provably Prioritizes a Shortcut Spurious Feature in the XOR Model
  32. NeurIPS 2026Shuffle and Joint Differential Privacy for Generalized Linear Contextual Bandits
  33. NeurIPS 2026Simple Extensions of Single-Objective Acquisition Functions and Hedge Strategies for Multi-Objective Bayesian Optimization
  34. NeurIPS 2026Simulating Students or Sycophantic Problem Solving? On Misconception Faithfulness of LLM Simulators
  35. NeurIPS 2026Sink vs. diagonal patterns as mechanisms for attention switch and oversmoothing prevention
  36. NeurIPS 2026SLAyiNG: A Diverse and Community-validated Dataset of Queer Slang
  37. NeurIPS 2026SLDR: Defending Against Malicious Fine-tuning via Selective Layers Recovery and Dynamic Routing
  38. NeurIPS 2026SoccerNarrate: Event-Grounded Streaming Soccer Commentary with Macro-Window Preference Alignment
  39. NeurIPS 2026Soft Token Alignment for Cross-Lingual Reasoning
  40. NeurIPS 2026Solving Stochastic Control under Multiplicative and Internal Noise via Constrained Optimization
  41. NeurIPS 2026SPACE: Unifying Symmetric and Asymmetric Routing Problems for Generalist Neural Solver
  42. NeurIPS 2026Sparsifying Correlation Clustering: Edge Coresets, Triangle Witnesses, and Observation Lower Bounds
  43. NeurIPS 2026Spectral Alignment in Forward–Backward Representations via Temporal Abstraction
  44. NeurIPS 2026Speculative Decoding Is Per-Token, Not Path
  45. NeurIPS 2026Spin-Weighted Spherical Harmonics Enable Complete and Scalable E(3)-Equivariant Networks
  46. NeurIPS 2026Sponsored Questions and How to Auction Them
  47. NeurIPS 2026Stability and Diversity of Networked Self-Consuming Generative Ecosystems
  48. NeurIPS 2026StaDy: Factorizing the World into Static and Dynamic via Likelihood Matching
  49. NeurIPS 2026Stage-Aware Dual Alignment for Covariate Shift in Graph Domain Adaptation
  50. NeurIPS 2026SteadyThought: Mitigating LLM Under-Thinking via Thought-Level Preference 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