社会层面 · safety论文趋势 2023–2026

主要结论

• 社会层面 · safety论文从 2023 年的 4 篇增长到 2026 年的 370 篇，三年间扩大到约 93 倍。
• ICML 是发社会层面 · safety论文最多的会议，2023–2026 年共 364 篇，占四个会议总量的 53%。
• 交叉方向里涨得最快的是对齐与安全：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 137 篇。
• 在「对齐与负责任 AI」的 5 个主题里，它的 2026 年论文数排第 3；同组增长最快的社会层面 · privacy是 6.6 倍，社会层面 · safety是 93 倍。

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

• NeurIPS 2026BarrierSteer: LLM Safety via Learning Barrier Steering
• ICML 2026Anti-Backdoor Coreset Selection via Cumulative Entropy
• ICLR 2026Bi-Criteria Metric Distortion
• CVPR 2026Omni-MMSI Toward Identity-attributed Social Interaction Understanding
• NeurIPS 2025Safety Pretraining: Toward the Next Generation of Safe AI
• ICML 2025$S^2$FGL: Spatial Spectral Federated Graph Learning
• ICLR 2025The Hidden Cost of Waiting for Accurate Predictions
• CVPR 2025Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models?
• NeurIPS 2024Quantifying the Bitter Lesson: How Safety Benchmarks Measure Capabilities Instead of Safety
• ICML 2024Optimally Improving Cooperative Learning in a Social Setting
• ICLR 2024The Devil is in the Neurons: Interpreting and Mitigating Social Biases in Language Models
• CVPR 2024Digital Life Project: Autonomous 3D Characters with Social Intelligence
• NeurIPS 2023Strategyproof Voting under Correlated Beliefs
• ICML 2023Adaptive Coordination in Social Embodied Rearrangement
• NeurIPS 2026Beyond Feature Disruption: Boundary-Diverting Unlearnable Examples against Linear Probing
• ICML 2026AutoControl Arena: Synthesizing Executable Test Environments for Frontier AI Risk Evaluation
• ICLR 2026PRISON: Unmasking the Criminal Potential of Large Language Models
• CVPR 2026Spot The Ball A Benchmark for Visual Social Inference
• NeurIPS 2025SAFEx: Analyzing Vulnerabilities of MoE-Based LLMs via Stable Safety-critical Expert Identification
• ICML 2025Adversaries Can Misuse Combinations of Safe Models

全部社会层面 · safety论文

共 688 篇，第 10 / 14 页

451. ICML 2025Not All Wrong is Bad: Using Adversarial Examples for Unlearning
452. ICML 2025On Explaining Equivariant Graph Networks via Improved Relevance Propagation
453. ICML 2025On the Impact of Performative Risk Minimization for Binary Random Variables
454. ICML 2025On the Vulnerability of Applying Retrieval-Augmented Generation within Knowledge-Intensive Application Domains
455. ICML 2025Optimizing Adaptive Attacks against Watermarks for Language Models
456. ICML 2025OR-Bench: An Over-Refusal Benchmark for Large Language Models
457. ICML 2025PASS: Private Attributes Protection with Stochastic Data Substitution
458. ICML 2025Pointwise Information Measures as Confidence Estimators in Deep Neural Networks: A Comparative Study
459. ICML 2025POROver: Improving Safety and Reducing Overrefusal in Large Language Models with Overgeneration and Preference Optimization
460. ICML 2025Position: AI Safety Must Embrace an Antifragile Perspective
461. ICML 2025Position: AI Safety should prioritize the Future of Work
462. ICML 2025Position: Editing Large Language Models Poses Serious Safety Risks
463. ICML 2025Position: Generative AI Regulation Can Learn from Social Media Regulation
464. ICML 2025Position: Humanity Faces Existential Risk from Gradual Disempowerment
465. ICML 2025Position: Political Neutrality in AI Is Impossible — But Here Is How to Approximate It
466. ICML 2025Position: Strong Consumer Protection is an Inalienable Defense for AI Safety in the United States
467. ICML 2025Position: The Right to AI
468. ICML 2025Position: We Need Responsible, Application-Driven (RAD) AI Research
469. ICML 2025Prediction via Shapley Value Regression
470. ICML 2025Preference learning made easy: Everything should be understood through win rate
471. ICML 2025Provably Cost-Sensitive Adversarial Defense via Randomized Smoothing
472. ICML 2025Ranked from Within: Ranking Large Multimodal Models Without Labels
473. ICML 2025Relative Error Fair Clustering in the Weak-Strong Oracle Model
474. ICML 2025Representative Ranking for Deliberation in the Public Sphere
475. ICML 2025Rethinking the Temperature for Federated Heterogeneous Distillation
476. ICML 2025SafeArena: Evaluating the Safety of Autonomous Web Agents
477. ICML 2025Safety Alignment Can Be Not Superficial With Explicit Safety Signals
478. ICML 2025Safety Reasoning with Guidelines
479. ICML 2025SafetyAnalyst: Interpretable, Transparent, and Steerable Safety Moderation for AI Behavior
480. ICML 2025Scaling Trends in Language Model Robustness
481. ICML 2025Selective Preference Aggregation
482. ICML 2025SEMU: Singular Value Decomposition for Efficient Machine Unlearning
483. ICML 2025ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning
484. ICML 2025Speak Easy: Eliciting Harmful Jailbreaks from LLMs with Simple Interactions
485. ICML 2025SPRI: Aligning Large Language Models with Context-Situated Principles
486. ICML 2025STAIR: Improving Safety Alignment with Introspective Reasoning
487. ICML 2025Statistical Collusion by Collectives on Learning Platforms
488. ICML 2025STD-FD: Spatio-Temporal Distribution Fitting Deviation for AIGC Forgery Identification
489. ICML 2025Stealix: Model Stealing via Prompt Evolution
490. ICML 2025Suitability Filter: A Statistical Framework for Classifier Evaluation in Real-World Deployment Settings
491. ICML 2025Sum-of-Parts: Self-Attributing Neural Networks with End-to-End Learning of Feature Groups
492. ICML 2025Tensor Product Neural Networks for Functional ANOVA Model
493. ICML 2025Test-Time Multimodal Backdoor Detection by Contrastive Prompting
494. ICML 2025The Elicitation Game: Evaluating Capability Elicitation Techniques
495. ICML 2025The Illusion of Role Separation: Hidden Shortcuts in LLM Role Learning (and How to Fix Them)
496. ICML 2025The Jailbreak Tax: How Useful are Your Jailbreak Outputs?
497. ICML 2025The Limits of Predicting Agents from Behaviour
498. ICML 2025The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
499. ICML 2025TOPLOC: A Locality Sensitive Hashing Scheme for Trustless Verifiable Inference
500. ICML 2025Towards Attributions of Input Variables in a Coalition

常见问题

2023–2026 年四大会议一共有多少篇社会层面 · safety论文？

共 688 篇，按年份分别是 2023 年 4 篇、2024 年 41 篇、2025 年 273 篇、2026 年 370 篇。

NeurIPS 2026 有多少篇社会层面 · safety论文？

NeurIPS 2026 有 121 篇社会层面 · safety论文，上一届是 131 篇。

ICLR 2026 有多少篇社会层面 · safety论文？

ICLR 2026 有 20 篇社会层面 · safety论文，上一届是 11 篇。

ICML 2026 有多少篇社会层面 · safety论文？

ICML 2026 有 226 篇社会层面 · safety论文，上一届是 130 篇。

CVPR 2026 有多少篇社会层面 · safety论文？

CVPR 2026 有 3 篇社会层面 · safety论文，上一届是 1 篇。

哪个会议的社会层面 · safety论文最多？

ICML，2023–2026 年共 364 篇。