mlsanityData to 14 Oct 2026

Watermarking Diffusion Language Models

arXiv 2509.24368 ↗First posted 29 Sep 2025cs.LG cs.AI cs.CRcited by 10

On the map: Diffusion models → Robustness, privacy and adversarial aspects

also in 1 more
Abstract (arXiv)

We introduce the first watermark tailored for diffusion language models (DLMs), an emergent LLM paradigm able to generate tokens in arbitrary order, in contrast to standard autoregressive language models (ARLMs) which generate tokens sequentially. While there has been much work in ARLM watermarking, a key challenge when attempting to apply these schemes directly to the DLM setting is that they rely on previously generated tokens, which are not always available with DLM generation. In this work we address this challenge by: (i) applying the watermark in expectation over the context even when some context tokens are yet to be determined, and (ii) promoting tokens which increase the watermark strength when used as context for other tokens. This is accomplished while keeping the watermark detector unchanged. Our experimental evaluation demonstrates that the DLM watermark leads to a >99% true positive rate with minimal quality impact and achieves similar robustness to existing ARLM watermarks, enabling for the first time reliable DLM watermarking.

Key facts. Cited by 10 papers in the map (data through 14 Oct 2026). Placed in Robustness, privacy and adversarial aspects, one of 14 directions in Diffusion models. Builds on 57 of its 68 references that are in the map. First posted on arXiv 29 Sep 2025.

Builds on

57 of 68 references.

Closest ancestors · 4

Same direction, newest first

  1. 2024Gaussian Shading: Provable Performance-Lossless Image Watermarking for Diffusion Models7 Apr
  2. 2024Jailbreaking Prompt Attack: A Controllable Adversarial Attack against Diffusion Models2 Apr
  3. 2023Tree-Ring Watermarks: Fingerprints for Diffusion Images that are Invisible and Robust31 May
  4. 2023The Stable Signature: Rooting Watermarks in Latent Diffusion Models27 Mar

Other references · 53

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2020Language Models are Few-Shot Learnerscited by 32,419
  3. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  4. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  5. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  6. 2020Denoising Diffusion Implicit Modelscited by 9,781
  7. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
  8. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  9. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  10. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  11. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,761
  12. 2021Program Synthesis with Large Language Modelscited by 3,637
38 more
  1. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,879
  2. 2023Sparks of Artificial General Intelligence: Early experiments with GPT-4cited by 2,592
  3. 2021Structured Denoising Diffusion Models in Discrete State-Spacescited by 1,719
  4. 2025Large Language Diffusion Modelscited by 912
  5. 2024Simple and Effective Masked Diffusion Language Modelscited by 848
  6. 2023AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedbackcited by 748
  7. 2023DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvaturecited by 726
  8. 2023A Watermark for Large Language Modelscited by 725
  9. 2024Simplified and Generalized Masked Diffusion for Discrete Datacited by 626
  10. 2019ELI5: Long Form Question Answeringcited by 575
  11. 2021FinQA: A Dataset of Numerical Reasoning over Financial Datacited by 562
  12. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Modelscited by 409
  13. 2023The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"cited by 407
  14. 2024Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Datacited by 381
  15. 2024Better & Faster Large Language Models via Multi-token Predictioncited by 360
  16. 2023Robust Distortion-free Watermarks for Language Modelscited by 320
  17. 2023Provable Robust Watermarking for AI-Generated Textcited by 299
  18. 2025MMaDA: Multimodal Large Diffusion Language Modelscited by 250
  19. 2025Mercury: Ultra-Fast Language Models Based on Diffusioncited by 235
  20. 2025Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusionscited by 214
  21. 2015Semi-supervised Sequence Learningcited by 198
  22. 2023On the Reliability of Watermarks for Large Language Modelscited by 189
  23. 2025LLaDA-V: Large Language Diffusion Models with Visual Instruction Tuningcited by 148
  24. 2025The Diffusion Dualitycited by 135
  25. 2024MarkLLM: An Open-Source Toolkit for LLM Watermarkingcited by 95
  26. 2024Watermark Stealing in Large Language Modelscited by 92
  27. 2023Reflected Diffusion Modelscited by 87
  28. 2023Three Bricks to Consolidate Watermarks for Large Language Modelscited by 84
  29. 2023WaterBench: Towards Holistic Evaluation of Watermarks for Large Language Modelscited by 32
  30. 2024Black-Box Detection of Language Model Watermarkscited by 23
  31. 2025Mark Your LLM: Detecting the Misuse of Open-Source Large Language Models via Watermarkingcited by 19
  32. 2025Watermarking Autoregressive Image Generationcited by 16
  33. 2024Large Language Model Watermark Stealing With Mixed Integer Programmingcited by 14
  34. 2024A Watermark for Order-Agnostic Language Modelscited by 14
  35. 2024Zero-Shot Detection of LLM-Generated Text using Token Cohesivenesscited by 13
  36. 2024Can Watermarked LLMs be Identified by Users via Crafted Prompts?cited by 13
  37. 2024Provably Robust Watermarks for Open-Source Language Modelscited by 13
  38. 2025GaussMark: A Practical Approach for Structural Watermarking of Language Modelscited by 13

50 most cited of 53 All 53 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 8.

10 papers.

  1. 2025DITTO: A Spoofing Attack Framework on Watermarked LLMs via Knowledge Distillationcited by 6
  2. 2025Watermarking Discrete Diffusion Language Modelscited by 5
  3. 2026LR-DWM: Efficient Watermarking for Diffusion Language Modelscited by 4
  4. 2026Chainwash: Multi-Step Rewriting Attacks on Diffusion Language Model Watermarkscited by 1
  5. 2026Every Bit, Everywhere, All at Once: A Binomial Multibit LLM Watermarkcited by 1
  6. 2026Global Sketch-Based Watermarking for Diffusion Language Modelscited by 1
  7. 2026Watermark Forensics for Generative Models: An Information-Theoretic Perspectivecited by 1
  8. 2026DenMark: Robust Semantic Watermarking for Diffusion Language Modelscited by 1
  9. 2026WoE Wrote It? Watermarking Mixture-of-Experts LLMs for Black-Box Text Provenancecited by 0
  10. 2026TANGO: Watermarking Masked Diffusion Language Models in Token Pairscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025DMark: Order-Agnostic Watermarking for Diffusion Large Language Models
  2. 2026dgMARK: Decoding-Guided Watermarking for Diffusion Language Models
  3. 2026WMVLM: Evaluating Diffusion Model Image Watermarking via Vision-Language Models
  4. 2024Theoretically Grounded Framework for LLM Watermarking: A Distribution-Adaptive Approach
  5. 2025HeavyWater and SimplexWater: Distortion-Free LLM Watermarks for Low-Entropy Next-Token Predictions
  6. 2023A Recipe for Watermarking Diffusion Models
  7. 2023Baselines for Identifying Watermarked Large Language Models
  8. 2024PostMark: A Robust Blackbox Watermark for Large Language Models