mlsanityData to 7 Oct 2026

MC2Mark: Distortion-Free Multi-Bit Watermarking for Long Messages

arXiv 2602.14030 ↗First posted 15 Feb 2026cs.CR cs.LGcited by 2

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

Abstract (arXiv)

Large language models now produce text indistinguishable from human writing, which increases the need for reliable provenance tracing. Multi-bit watermarking can embed identifiers into generated text, but existing methods struggle to keep both text quality and watermark strength while carrying long messages. We propose MC2Mark, a distortion-free multi-bit watermarking framework designed for reliable embedding and decoding of long messages. Our key technical idea is Multi-Channel Colored Reweighting, which encodes bits through structured token reweighting while keeping the token distribution unbiased, together with Multi-Layer Sequential Reweighting to strengthen the watermark signal and an evidence-accumulation detector for message recovery. Experiments show that MC2Mark improves detectability and robustness over prior multi-bit watermarking methods while preserving generation quality, achieving near-perfect accuracy for short messages and exceeding the second-best method by nearly 30% for long messages.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Watermarking generated content and models, one of 12 directions in LLM safety and security. Builds on 24 of its 28 references that are in the map. First posted on arXiv 15 Feb 2026.

Builds on

24 of 28 references.

Closest ancestors · 17

Same direction, newest first

  1. 2025An Ensemble Framework for Unbiased Language Model Watermarking28 Sep
  2. 2025BiMark: Unbiased Multilayer Watermarking for Large Language Models19 Jun
  3. 2025Improved Unbiased Watermark for Large Language Models16 Feb
  4. 2024Provably Robust Multi-bit Watermarking for AI-generated Text30 Jan
  5. 2023Mark My Words: Analyzing and Evaluating Language Model Watermarks1 Dec
  6. 2023Publicly-Detectable Watermarking for Language Models27 Oct
  7. 2023A Resilient and Accessible Distribution-Preserving Watermark for Large Language Models11 Oct
  8. 2023A Semantic Invariant Robust Watermark for Large Language Models10 Oct
  9. 2023Unbiased Watermark for Large Language Models22 Sep
  10. 2023Advancing Beyond Identification: Multi-bit Watermark for Large Language Models1 Aug
  11. 2023An Unforgeable Publicly Verifiable Watermark for Large Language Models30 Jul
  12. 2023Robust Distortion-free Watermarks for Language Models28 Jul
5 more
  1. 2023Three Bricks to Consolidate Watermarks for Large Language Models26 Jul
  2. 2023Provable Robust Watermarking for AI-Generated Text30 Jun
  3. 2023Undetectable Watermarks for Language Models25 May
  4. 2023Paraphrasing evades detectors of AI-generated text, but retrieval is an effective defense23 Mar
  5. 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 14,491
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2019BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehensioncited by 5,874
  4. 2024Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phonecited by 1,883
  5. 2020Multilingual Denoising Pre-training for Neural Machine Translationcited by 1,028
  6. 2021Ethical and social risks of harm from Language Modelscited by 957
  7. 2023Generative Language Models and Automated Influence Operations: Emerging Threats and Potential Mitigationscited by 157

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Every Bit, Everywhere, All at Once: A Binomial Multibit LLM Watermarkcited by 1
  2. 2026Learning to Follow In-Context Watermark Instructions via Self-Distillationcited by 0