Lune

ACL Findings2024

Duwak: Dual Watermarks in Large Language Models

Chaoyi Zhu, Jeroen Galjaard, Pin-Yu Chen, Lydia Y. Chen

Abstract

As large language models (LLM) are increasingly used for text generation tasks, it is critical to audit their usages, govern their applications, and mitigate their potential harms. Existing watermark techniques are shown effective in embedding single human-imperceptible and machine-detectable patterns without significantly affecting generated text quality and semantics. However, the efficiency in detecting watermarks, i.e., the minimum number of tokens required to assert detection with significance and robustness against post-editing, is still debatable. In this paper, we propose, Duwak, to fundamentally enhance the efficiency and quality of watermarking by embedding dual secret patterns in both token probability distribution and sampling schemes. To mitigate expression degradation caused by biasing toward specific tokens, we design a contrastive search to watermark the sampling scheme, which minimizes the token repetition and enhances the diversity. We theoretically explain the interdependency of the two watermarks within Duwak. We evaluate Duwak extensively on Llama2 and Vicuna under various post-editing attacks, against four state-of-theart watermarking techniques and combinations of them. Our results show that Duwak marked text achieves the highest watermarked text quality at the lowest required token count for detection, up to 70% tokens less than existing approaches, especially under post paraphrasing. Our code is available at https://github. com/chaoyitud/Dual-Watermarks.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

6 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Neural Text Generation With Unlikelihood Training cites The Curious Case of Neural Text Degeneration
  • The Curious Case of Neural Text Degeneration cites Neural Text Generation With Unlikelihood Training
  • Ensemble Watermarks for Large Language Models cites DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
  • Ensemble Watermarks for Large Language Models cites A Watermark for Large Language Models
  • MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers cites Neural Text Generation With Unlikelihood Training
  • MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers cites The Curious Case of Neural Text Degeneration
  • DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature cites A Watermark for Large Language Models
  • TabWak: A Watermark for Tabular Diffusion Models cites A Watermark for Large Language Models
  • WaterMax: breaking the LLM watermark detectability-robustness-quality trade-off cites MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers
  • WaterMax: breaking the LLM watermark detectability-robustness-quality trade-off cites DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvature
  • WaterMax: breaking the LLM watermark detectability-robustness-quality trade-off cites A Watermark for Large Language Models
  • From Trade-off to Synergy: A Versatile Symbiotic Watermarking Framework for Large Language Models cites A Watermark for Large Language Models
  • TimeWak: Temporal Chained-Hashing Watermark for Time Series Data cites TabWak: A Watermark for Tabular Diffusion Models
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers6

  1. WaterMax: breaking the LLM watermark detectability-robustness-quality trade-offEva Giboulot, Teddy FuronNeurIPS 2024 · 76 citations
  2. TimeWak: Temporal Chained-Hashing Watermark for Time Series DataZhi Wen Soi, Chaoyi Zhu, Fouad Abiad, Aditya Shankar et al.NeurIPS 2025 · 5 citations
  3. From Trade-off to Synergy: A Versatile Symbiotic Watermarking Framework for Large Language ModelsYidan Wang, Yubing Ren, Yanan Cao, Binxing FangACL 2025 · 4 citations
  4. Ensemble Watermarks for Large Language ModelsGeorg Niess, Roman KernACL 2025 · 3 citations
  5. TabWak: A Watermark for Tabular Diffusion ModelsChaoyi Zhu, Jiayi Tang, Jeroen M. Galjaard, Pin-Yu Chen et al.ICLR 2025

Builds on9

  1. The Curious Case of Neural Text DegenerationAri Holtzman, Jan Buys, Li Du, Maxwell Forbes et al.ICLR 2020 · 4,112 citations
  2. DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability CurvatureEric Mitchell, Yoonho Lee, Alexander Khazatsky, Christopher D. Manning et al.ICML 2023 · 988 citations
  3. A Watermark for Large Language ModelsJohn Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz et al.ICML 2023 · 854 citations
  4. Neural Text Generation With Unlikelihood TrainingSean Welleck, Ilia Kulikov, Stephen Roller, Emily Dinan et al.ICLR 2020 · 683 citations
  5. MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence FrontiersKrishna Pillutla, Swabha Swayamdipta, Rowan Zellers, John Thickstun et al.NeurIPS 2021 · 606 citations

Related papers

  1. Provable Robust Watermarking for AI-Generated TextXuandong Zhao, Prabhanjan Vijendra Ananth, Lei Li, Yu-Xiang WangICLR 2024 · 312 citations
  2. Subtle Signatures, Strong Shields: Advancing Robust and Imperceptible Watermarking in Large Language ModelsYubing Ren, Ping Guo, Yanan Cao, Wei MaACL Findings 2024 · 6 citations
  3. Catch-22: On the Fundamental Tradeoff Between Detectability and Robustness in LLM WatermarkingKuheli Pratihar, Debdeep MukhopadhyayICML 2026
  4. Theoretically Grounded Framework for LLM Watermarking: A Distribution-Adaptive ApproachHaiyun He, Yepeng Liu, Ziqiao Wang, Yongyi Mao et al.NeurIPS 2025 · 26 citations
  5. On the Reliability of Watermarks for Large Language ModelsJohn Kirchenbauer, Jonas Geiping, Yuxin Wen, Manli Shu et al.ICLR 2024 · 202 citations
Dusk over the sea between two cliffs drawn in fine vertical lines