mlsanityData to 14 Oct 2026

WaterMax: breaking the LLM watermark detectability-robustness-quality trade-off

arXiv 2403.04808 ↗First posted 6 Mar 2024cs.CR cs.CL cs.LGcited by 54

On the map: LLM safety and security → Watermarking generated content and models

Abstract (arXiv)

Watermarking is a technical means to dissuade malfeasant usage of Large Language Models. This paper proposes a novel watermarking scheme, so-called WaterMax, that enjoys high detectability while sustaining the quality of the generated text of the original LLM. Its new design leaves the LLM untouched (no modification of the weights, logits, temperature, or sampling technique). WaterMax balances robustness and complexity contrary to the watermarking techniques of the literature inherently provoking a trade-off between quality and robustness. Its performance is both theoretically proven and experimentally validated. It outperforms all the SotA techniques under the most complete benchmark suite. Code available at https://github.com/eva-giboulot/WaterMax.

Key facts. Cited by 54 papers in the map (data through 14 Oct 2026). Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security. Builds on 20 of its 38 references that are in the map. First posted on arXiv 6 Mar 2024.

Builds on

20 of 38 references.

Closest ancestors · 13

Same direction, newest first

  1. 2024Token-Specific Watermarking with Enhanced Detectability and Semantic Coherence for Large Language Models28 Feb
  2. 2024Can Watermarks Survive Translation? On the Cross-lingual Consistency of Text Watermark for Large Language Models21 Feb
  3. 2024Adaptive Text Watermark for Large Language Models25 Jan
  4. 2023On the Learnability of Watermarks for Language Models7 Dec
  5. 2023Mark My Words: Analyzing and Evaluating Language Model Watermarks1 Dec
  6. 2023Publicly-Detectable Watermarking for Language Models27 Oct
  7. 2023Unbiased Watermark for Large Language Models22 Sep
  8. 2023Robust Distortion-free Watermarks for Language Models28 Jul
  9. 2023Three Bricks to Consolidate Watermarks for Large Language Models26 Jul
  10. 2023Provable Robust Watermarking for AI-Generated Text30 Jun
  11. 2023On the Reliability of Watermarks for Large Language Models7 Jun
  12. 2023Undetectable Watermarks for Language Models25 May
1 more
  1. 2023A Watermark for Large Language Models24 Jan

Other references · 7

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  2. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  3. 2022OPT: Open Pre-trained Transformer Language Modelscited by 3,512
  4. 2021MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontierscited by 470
  5. 2024Spotting LLMs With Binoculars: Zero-Shot Detection of Machine-Generated Textcited by 263
  6. 2023LLMDet: A Third Party Large Language Models Generated Text Detection Toolcited by 45
  7. 2021PPL-MCTS: Constrained Textual Generation Through Discriminator-Guided MCTS Decodingcited by 33

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 9; 2025: 25; 2026: 20.

54 papers.

Sort:

By area:

  1. 2024SoK: Watermarking for AI-Generated Contentcited by 51
  2. 2025MorphMark: Flexible Adaptive Watermarking for Large Language Modelscited by 24
  3. 2024Watermarking Techniques for Large Language Models: A Surveycited by 23
  4. 2024Theoretically Grounded Framework for LLM Watermarking: A Distribution-Adaptive Approachcited by 22
  5. 2025AdaDetectGPT: Adaptive Detection of LLM-Generated Text with Statistical Guaranteescited by 21
  6. 2025CoheMark: A Novel Sentence-Level Watermark for Enhanced Text Qualitycited by 19
  7. 2024A Watermark for Black-Box Language Modelscited by 16
  8. 2025PMark: Towards Robust and Distortion-free Semantic-level Watermarking with Channel Constraintscited by 16
  9. 2024SilverSpeak: Evading AI-Generated Text Detectors using Homoglyphscited by 13
  10. 2024WAPITI: A Watermark for Finetuned Open-Source LLMscited by 12
  11. 2025Dataset Protection via Watermarked Canaries in Retrieval-Augmented LLMscited by 12
  12. 2025Marking Code Without Breaking It: Code Watermarking for Detecting LLM-Generated Codecited by 10
38 more
  1. 2025In-Context Watermarks for Large Language Modelscited by 10
  2. 2024Watermark under Fire: A Robustness Evaluation of LLM Watermarkingcited by 9
  3. 2025Enhancing LLM Watermark Resilience Against Both Scrubbing and Spoofing Attackscited by 9
  4. 2026Learn-to-Distance: Distance Learning for Detecting LLM-Generated Textcited by 8
  5. 2025Position: LLM Watermarking Should Align Stakeholders' Incentives for Practical Adoptioncited by 7
  6. 2026PASA: A Principled Embedding-Space Watermarking Approach for LLM-Generated Text under Semantic-Invariant Attackscited by 7
  7. 2024WaterPool: A Watermark Mitigating Trade-offs among Imperceptibility, Efficacy and Robustnesscited by 6
  8. 2025WaterMod: Modular Token-Rank Partitioning for Probability-Balanced LLM Watermarkingcited by 6
  9. 2024Survey on Plagiarism Detection in Large Language Models: The Impact of ChatGPT and Gemini on Academic Integritycited by 5
  10. 2025On the Empirical Power of Goodness-of-Fit Tests in Watermark Detectioncited by 5
  11. 2025Watermarking Discrete Diffusion Language Modelscited by 5
  12. 2026TextSeal: A Localized LLM Watermark for Provenance & Distillation Protectioncited by 5
  13. 2025Majority Bit-Aware Watermarking For Large Language Modelscited by 4
  14. 2025Watermarking Degrades Alignment in Language Models: Analysis and Mitigationcited by 3
  15. 2025Multi-use LLM Watermarking and the False Detection Problemcited by 3
  16. 2025Detecting Post-generation Edits to Watermarked LLM Outputs via Combinatorial Watermarkingcited by 3
  17. 2025δ-STEAL: LLM Stealing Attack with Local Differential Privacycited by 3
  18. 2026Distilling the Thought, Watermarking the Answer: A Principle Semantic Guided Watermark for Large Reasoning Modelscited by 3
  19. 2026XMark: Reliable Multi-Bit Watermarking for LLM-Generated Textscited by 3
  20. 2025SoK: Are Watermarks in LLMs Ready for Deployment?cited by 2
  21. 2025SimKey: A Semantically Aware Key Module for Watermarking Language Modelscited by 2
  22. 2025A Linguistics-Aware LLM Watermarking via Syntactic Predictabilitycited by 2
  23. 2025How Good is Post-Hoc Watermarking With Language Model Rephrasing?cited by 2
  24. 2026More Haste, Less Speed: Weaker Single-Layer Watermark Improves Distortion-Free Watermark Ensemblescited by 2
  25. 2026SLAM: Structural Linguistic Activation Marking for Language Modelscited by 2
  26. 2026SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustnesscited by 2
  27. 2025MUSE: Model-Agnostic Tabular Watermarking via Multi-Sample Selectioncited by 1
  28. 2025HeavyWater and SimplexWater: Distortion-Free LLM Watermarks for Low-Entropy Next-Token Predictionscited by 1
  29. 2026Improving the Trade-off Between Watermark Strength and Speculative Sampling Efficiency for Language Modelscited by 1
  30. 2026SKeDA: A Generative Watermarking Framework for Text-to-video Diffusion Modelscited by 1
  31. 2026Robust LLM Watermarking with Minimal Semantic Distortion for IP Protectioncited by 1
  32. 2026LLM Watermarking as Big Data Provenance: A Deployment-Oriented Systematizationcited by 1
  33. 2026SAC-Copula: Quality-Preserving Watermarking for Diffusion Language Models via Smooth Correlated Gumbel Fieldscited by 1
  34. 2026Semantic Watermarking with Order-Robust Detection over Sub-sentence Unitscited by 1
  35. 2025Improving Detection of Watermarked Language Modelscited by 0
  36. 2025WaterSearch: Exploring Seed Pooling for Improving the Quality-Detectability Trade-off in LLM Watermarkingcited by 0
  37. 2026Trustworthy AI: Ensuring Reliability and Accountability from Models to Agentscited by 0
  38. 2026DHMark: Public-Key Watermarking for LLM-Generated Text via Diffie-Hellman-Guided Rejection Samplingcited by 0

50 most cited of 54 All 54 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2024Black-Box Detection of Language Model Watermarks
  2. 2024Less is More: Sparse Watermarking in LLMs with Enhanced Text Quality
  3. 2025An End-to-End Model for Logits-Based Large Language Models Watermarking
  4. 2025PRO: Enabling Precise and Robust Text Watermark for Open-Source LLMs
  5. 2025Towards Watermarking of Open-Source LLMs
  6. 2024Optimizing Adaptive Attacks against Watermarks for Language Models
  7. 2023New Evaluation Metrics Capture Quality Degradation due to LLM Watermarking
  8. 2024Can Watermarked LLMs be Identified by Users via Crafted Prompts?