{
  "abstract": "We study the problem of multi-bit watermarking for non-autoregressive large language models (LLMs). We introduce an information-theoretic model inspired by diffusion language models (DLM), in which the encoder has limited non-causal access to token distributions within each token block. This formulation enables an information-theoretic characterization of the non-causal watermarking capacity, in which knowledge of LLM cover statistics is leveraged to enable a multi-bit covert embedding. We study the information-theoretic limits of the model by combining Gelfand--Pinsker and channel synthesis coding techniques and obtain an exact characterization of the capacity. The embedding strategy is further optimized across blocks using a constrained Markov decision process (CMDP) and we develop an explicit algorithm based on polar codes following the information-theoretic principles. We simulate the error performance on LLaDA, and provide empirical total variation (TV) analysis as a function of key randomness.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2605.16709",
  "author_total": 4,
  "authors": [
    {
      "name": "Sidong Guo"
    },
    {
      "name": "Tyler Kann"
    },
    {
      "name": "Teodora Baluta"
    },
    {
      "name": "Matthieu R. Bloch"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/9mv7fmpebr3r7.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2605.16709",
  "fulltext_url": null,
  "id": "9mv7fmpebr3r7",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2605.16709",
  "language": null,
  "licence": "CC-BY-4.0",
  "licence_label": "CC BY 4.0",
  "licence_source": "article",
  "licence_url": "http://creativecommons.org/licenses/by/4.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.IT"
    ],
    [
      "arxiv",
      "math.IT"
    ],
    [
      "openalex",
      "1707|Advanced Steganography and Watermarking Techniques"
    ]
  ],
  "openalex": "W7161725420",
  "pdf_url": "https://arxiv.org/pdf/2605.16709",
  "pmcid": null,
  "pmid": null,
  "published": "2026-09-02",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/9mv7fmpebr3r7.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:18:20Z",
      "native_id": "2605.16709",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2605.16709"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:math.IT",
      "subfield": "general"
    }
  ],
  "title": "Covert Multi-bit LLM Watermarking: An Information Theory and Coding Approach",
  "title_text": "Covert Multi-bit LLM Watermarking: An Information Theory and Coding Approach",
  "type": "preprint",
  "updated_at": "2026-10-30T12:18:20Z",
  "url": "https://research.draconitic.com/paper/9mv7fmpebr3r7",
  "venue": "arXiv",
  "versions": []
}