mlsanityData to 14 Oct 2026

Advancing Beyond Identification: Multi-bit Watermark for Large Language Models

arXiv 2308.00221 ↗First posted 1 Aug 2023cs.CL cs.AI cs.CRcited by 57

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

Abstract (arXiv)

We show the viability of tackling misuses of large language models beyond the identification of machine-generated text. While existing zero-bit watermark methods focus on detection only, some malicious misuses demand tracing the adversary user for counteracting them. To address this, we propose Multi-bit Watermark via Position Allocation, embedding traceable multi-bit information during language model generation. Through allocating tokens onto different parts of the messages, we embed longer messages in high corruption settings without added latency. By independently embedding sub-units of messages, the proposed method outperforms the existing works in terms of robustness and latency. Leveraging the benefits of zero-bit watermarking, our method enables robust extraction of the watermark without any model access, embedding and extraction of long messages (≥ 32-bit) without finetuning, and maintaining text quality, while allowing zero-bit detection all at the same time. Code is released here: https://github.com/bangawayoo/mb-lm-watermarking

Key facts. Cited by 57 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 28 of its 58 references that are in the map. First posted on arXiv 1 Aug 2023.

Builds on

28 of 58 references.

Closest ancestors · 12

Same direction, newest first

  1. 2023Robust Distortion-free Watermarks for Language Models28 Jul
  2. 2023Three Bricks to Consolidate Watermarks for Large Language Models26 Jul
  3. 2023On the Reliability of Watermarks for Large Language Models7 Jun
  4. 2023Undetectable Watermarks for Language Models25 May
  5. 2023Who Wrote this Code? Watermarking for Code Generation24 May
  6. 2023Robust Multi-bit Natural Language Watermarking through Invariant Features3 May
  7. 2023The Stable Signature: Rooting Watermarks in Latent Diffusion Models27 Mar
  8. 2023Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense23 Mar
  9. 2023A Recipe for Watermarking Diffusion Models17 Mar
  10. 2023A Watermark for Large Language Models24 Jan
  11. 2021Tracing Text Provenance via Context-Aware Lexical Substitution15 Dec
  12. 2021Protecting Intellectual Property of Language Generation APIs with Lexical Watermark5 Dec

Other references · 16

  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. 2016Pointer Sentinel Mixture Modelscited by 3,096
  5. 2023DetectGPT: Zero-Shot Machine-Generated Text Detection using Probability Curvaturecited by 726
  6. 2018HiDDeN: Hiding Data With Deep Networkscited by 397
  7. 2020Adversarial Watermarking Transformer: Towards Tracing Text Provenance with Data Hidingcited by 130
  8. 2023M4: Multi-generator, Multi-domain, and Multi-lingual Black-Box Machine-Generated Text Detectioncited by 121
  9. 2020Distortion Agnostic Deep Watermarkingcited by 90
  10. 2019Neural Linguistic Steganographycited by 78
  11. 2017Generating Steganographic Text with LSTMscited by 44
  12. 2022Perfectly Secure Steganography Using Minimum Entropy Couplingcited by 44
4 more
  1. 2018Analyzing the Digital Traces of Political Manipulation: The 2016 Russian Interference Twitter Campaigncited by 30
  2. 2021Paraphrastic Representations at Scalecited by 21
  3. 2022Propaganda and Misinformation on Facebook and Twitter during the Russian Invasion of Ukrainecited by 10
  4. 2022Binary Error-Correcting Codes with Minimal Noiseless Feedbackcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2023: 2; 2024: 8; 2025: 17; 2026: 30.

57 papers.

Sort:

By area:

  1. 2023A Survey of Text Watermarking in the Era of Large Language Modelscited by 112
  2. 2023On the Learnability of Watermarks for Language Modelscited by 81
  3. 2024Watermarking Makes Language Models Radioactivecited by 58
  4. 2024Provably Robust Multi-bit Watermarking for AI-generated Textcited by 47
  5. 2025Towards Trustworthy Retrieval Augmented Generation for Large Language Models: A Surveycited by 30
  6. 2024Detecting AI-Generated Text: Factors Influencing Detectability with Current Methodscited by 23
  7. 2024Watermarking Techniques for Large Language Models: A Surveycited by 23
  8. 2025BiMark: Unbiased Multilayer Watermarking for Large Language Modelscited by 21
  9. 2025StealthInk: A Multi-bit and Stealthy Watermark for Large Language Modelscited by 12
  10. 2025Distributional Information Embedding: A Framework for Multi-bit Watermarkingcited by 9
  11. 2025Enhancing LLM Watermark Resilience Against Both Scrubbing and Spoofing Attackscited by 9
  12. 2024Watermarking Language Models through Language Modelscited by 8
38 more
  1. 2025SAEMark: Steering Personalized Multilingual LLM Watermarks with Sparse Autoencoderscited by 7
  2. 2026SLICE: Semantic Latent Injection via Compartmentalized Embedding for Image Watermarkingcited by 7
  3. 2026TextSeal: A Localized LLM Watermark for Provenance & Distillation Protectioncited by 5
  4. 2026TRACE: A Two-Channel Robust Attribution Watermark via Complementary Embeddings for LLM-Agent Trajectoriescited by 5
  5. 2024CredID: Credible Multi-Bit Watermark for Large Language Models Identificationcited by 4
  6. 2025Majority Bit-Aware Watermarking For Large Language Modelscited by 4
  7. 2026MC2Mark: Distortion-Free Multi-Bit Watermarking for Long Messagescited by 4
  8. 2025BiMarker: Enhancing Text Watermark Detection for Large Language Models with Bipolar Watermarkscited by 3
  9. 2025Watermarking Degrades Alignment in Language Models: Analysis and Mitigationcited by 3
  10. 2025Enhancing Watermarking Quality for LLMs via Contextual Generation States Awarenesscited by 3
  11. 2025CATMark: A Context-Aware Thresholding Framework for Robust Cross-Task Watermarking in Large Language Modelscited by 3
  12. 2026Distilling the Thought, Watermarking the Answer: A Principle Semantic Guided Watermark for Large Reasoning Modelscited by 3
  13. 2026MirrorMark: Generalizable Mirrored Sampling for Multi-bit LLM Watermarkingcited by 3
  14. 2026ArcMark: Distortion-Free Multi-Byte LLM Watermark via Optimal Transportcited by 3
  15. 2025SoK: Are Watermarks in LLMs Ready for Deployment?cited by 2
  16. 2025How Good is Post-Hoc Watermarking With Language Model Rephrasing?cited by 2
  17. 2026Beyond Raw Detection Scores: Markov-Informed Calibration for Boosting Machine-Generated Text Detectioncited by 2
  18. 2026Undetectable Conversations Between AI Agents via Pseudorandom Noise-Resilient Key Exchangecited by 2
  19. 2026Block-wise Codeword Embedding for Reliable Multi-bit Text Watermarkingcited by 2
  20. 2026Watermarking Should Be Treated as a Monitoring Primitivecited by 2
  21. 2026SAMark: A Self-Anchored Text Watermarking with Paragraph-Level Paraphrase Robustnesscited by 2
  22. 2024TokenMark: A Modality-Agnostic Watermark for Pre-trained Transformerscited by 1
  23. 2025LLM-Text Watermarking based on Lagrange Interpolationcited by 1
  24. 2025HeavyWater and SimplexWater: Distortion-Free LLM Watermarks for Low-Entropy Next-Token Predictionscited by 1
  25. 2026Every Bit, Everywhere, All at Once: A Binomial Multibit LLM Watermarkcited by 1
  26. 2026Robust LLM Watermarking with Minimal Semantic Distortion for IP Protectioncited by 1
  27. 2026AliMark: Enhancing Robustness of Sentence-Level Watermarking Against Text Paraphrasingcited by 1
  28. 2026Geometric Reciprocity: Unlocking Self-Supervision for Stereoscopic Video Generationcited by 1
  29. 2026LLM Watermarking as Big Data Provenance: A Deployment-Oriented Systematizationcited by 1
  30. 2026Watermark Forensics for Generative Models: An Information-Theoretic Perspectivecited by 1
  31. 2024Learnable Linguistic Watermarks for Tracing Model Extraction Attacks on Large Language Modelscited by 0
  32. 2025Toward Copyright Integrity and Verifiability via Multi-Bit Watermarking for Intelligent Transportation Systemscited by 0
  33. 2025Modification and Generated-Text Detection: Achieving Dual Detection Capabilities for the Outputs of LLM by Watermarkcited by 0
  34. 2026WorldCup Sampling for Multi-bit LLM Watermarkingcited by 0
  35. 2026Optimal Multi-bit Generative Watermarking Schemes Under Worst-Case False-Alarm Constraintscited by 0
  36. 2026TimeMark: A Trustworthy Time Watermarking Framework for Exact Generation-Time Recovery from AIGCcited by 0
  37. 2026Trustworthy AI: Ensuring Reliability and Accountability from Models to Agentscited by 0
  38. 2026CORE-BREW: LLR-Based Soft Decoding for Robust Multi-Bit LLM Watermarkingcited by 0

50 most cited of 57 All 57 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025DERMARK: A Dynamic, Efficient and Robust Multi-bit Watermark for Large Language Models
  2. 2024Let Watermarks Speak: A Robust and Unforgeable Watermark for Language Models
  3. 2024Multi-Bit Distortion-Free Watermarking for Large Language Models
  4. 2026XMark: Reliable Multi-Bit Watermarking for LLM-Generated Texts
  5. 2024Watermarking Language Models for Many Adaptive Users
  6. 2023Baselines for Identifying Watermarked Large Language Models
  7. 2026CertMark: Distortion-Free Multi-Bit Watermarking with Certified Decoding
  8. 2024Provably Robust Watermarks for Open-Source Language Models