Salvato in:
Dettagli Bibliografici
Autori principali: Guo, Sidong, Kann, Tyler, Baluta, Teodora, Bloch, Matthieu R.
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.16709
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866916018081759232
author Guo, Sidong
Kann, Tyler
Baluta, Teodora
Bloch, Matthieu R.
author_facet Guo, Sidong
Kann, Tyler
Baluta, Teodora
Bloch, Matthieu R.
contents We study the problem of multi-bit watermarking for large language models (LLMs). We introduce a block-autoregressive model inspired by multi-token prediction, in which the encoder has limited non-causal access to token distributions within each block. This formulation enables an information-theoretic characterization of multi-bit watermarking capacity, by which the 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. Our algorithm achieves a bit-error rate below 10 percent with a rate of 0.375 bits/token over short token lengths with negligible perplexity and distortion degradation.
format Preprint
id arxiv_https___arxiv_org_abs_2605_16709
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Covert Multi-bit LLM Watermarking: An Information Theory and Coding Approach
Guo, Sidong
Kann, Tyler
Baluta, Teodora
Bloch, Matthieu R.
Information Theory
We study the problem of multi-bit watermarking for large language models (LLMs). We introduce a block-autoregressive model inspired by multi-token prediction, in which the encoder has limited non-causal access to token distributions within each block. This formulation enables an information-theoretic characterization of multi-bit watermarking capacity, by which the 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. Our algorithm achieves a bit-error rate below 10 percent with a rate of 0.375 bits/token over short token lengths with negligible perplexity and distortion degradation.
title Covert Multi-bit LLM Watermarking: An Information Theory and Coding Approach
topic Information Theory
url https://arxiv.org/abs/2605.16709