水印论文趋势 2023–2026

主要结论

• 水印论文从 2023 年的 6 篇增长到 2026 年的 56 篇，三年间扩大到约 9.3 倍。
• NeurIPS 是发水印论文最多的会议，2023–2026 年共 34 篇，占四个会议总量的 34%。
• 交叉方向里涨得最快的是社会层面 · accountability transparency and interpretability：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 18 篇。
• 在「安全与可信」的 5 个主题里，它的 2026 年论文数排第 3；同组增长最快的验证是 6.0 倍，水印是 9.3 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 2026Auditing Cross-Lingual Fairness in Language Model Watermarking
• ICML 2026Ideal Attribution and Faithful Watermarks for Language Models
• ICLR 2026An Ensemble Framework for Unbiased Language Model Watermarking
• CVPR 2026Convolutional Neural Networks Driven by Content Similarity
• NeurIPS 2025BitMark: Watermarking Bitwise Autoregressive Image Generative Models
• ICML 2025STAMP Your Content: Proving Dataset Membership via Watermarked Rephrasings
• ICLR 2025A Watermark for Order-Agnostic Language Models
• CVPR 2025OmniGuard: Hybrid Manipulation Localization via Augmented Versatile Deep Image Watermarking
• NeurIPS 2024FreqMark: Invisible Image Watermarking via Frequency Based Optimization in Latent Space
• ICML 2024DE-COP: Detecting Copyrighted Content in Language Models Training Data
• ICLR 2024On the Learnability of Watermarks for Language Models
• CVPR 2024EditGuard: Versatile Image Watermarking for Tamper Localization and Copyright Protection
• NeurIPS 2023An information-theoretic quantification of the content of communication between brain regions
• ICML 2023Margin-based Neural Network Watermarking
• CVPR 2023CAP-VSTNet: Content Affinity Preserved Versatile Style Transfer
• NeurIPS 2026Face Deepfake-aware Recovery via Semantic-driven Facial Representation-based Watermarking
• ICML 2026Towards Reliable Marking and Verification of AI-Generated Text via Geometry-aware Sentence-level Watermarking
• ICLR 2026Analyzing and Evaluating Unbiased Language Model Watermark
• CVPR 2026Decoupling Defense Strategies for Robust Image Watermarking
• NeurIPS 2025Provable Watermarking for Data Poisoning Attacks

全部水印论文

共 18 篇，第 1 / 1 页

1. NeurIPS 2026A Locally Tokenized Generative Model for Robust Time-Series Watermarking
2. NeurIPS 2026Asymmetric Phase Coding Audio Watermarking
3. NeurIPS 2026Auditing Cross-Lingual Fairness in Language Model Watermarking
4. NeurIPS 2026Face Deepfake-aware Recovery via Semantic-driven Facial Representation-based Watermarking
5. NeurIPS 2026Invisible Ink, Visible Lies: How Production Watermarking Causes LLMs to Hallucinate
6. NeurIPS 2026Majority Bit-Aware Watermarking for Large Language Models
7. NeurIPS 2026MarkTune: Improving the Quality-Detectability Trade-off in Model-Embedded LLM Watermarking
8. NeurIPS 2026MC$^2$Mark: Distortion-Free Multi-Bit Watermarking for Long Messages
9. NeurIPS 2026Multi-bit LLM Watermarking with Certified Semantic Distortion
10. NeurIPS 2026PP-Mark: Provable and Publicly Verifiable Watermarking for Generative AI
11. NeurIPS 2026Reading the Finetuning Prior: Verbatim Content Recovery via Contrastive Decoding Diffing
12. NeurIPS 2026Robust and Hard-to-Remove GNN Watermarking via Topological Invariant Perception
13. NeurIPS 2026Secure Seed-Based Multi-bit Watermarking for Diffusion Models from First Principles
14. NeurIPS 2026Sequential Behavioral Watermarking for LLM Agents
15. NeurIPS 2026TRACE: Structure-Aware Character Encoding for Robust and Generalizable Document Watermarking
16. NeurIPS 2026Watermarking as a Learned Intrinsic Property of Diffusion Models
17. NeurIPS 2026Watermarking Game-Playing Agents in Perfect-Information Extensive-Form Games
18. NeurIPS 2026Watermarking Without Standards Is Not AI Governance

常见问题

2023–2026 年四大会议一共有多少篇水印论文？

共 101 篇，按年份分别是 2023 年 6 篇、2024 年 17 篇、2025 年 22 篇、2026 年 56 篇。

NeurIPS 2026 有多少篇水印论文？

NeurIPS 2026 有 18 篇水印论文，上一届是 10 篇。

ICLR 2026 有多少篇水印论文？

ICLR 2026 有 11 篇水印论文，上一届是 2 篇。

ICML 2026 有多少篇水印论文？

ICML 2026 有 16 篇水印论文，上一届是 6 篇。

CVPR 2026 有多少篇水印论文？

CVPR 2026 有 11 篇水印论文，上一届是 4 篇。

哪个会议的水印论文最多？

NeurIPS，2023–2026 年共 34 篇。