@inproceedings{huang-etal-2026-revisiting,
title = "Revisiting Segmentation in Multi-bit {LLM} Watermarking",
author = "Huang, Beining and
Liu, Youdu and
Sun, Fei and
Su, Du and
Fu, Shuangjie and
Shen, Huawei and
Cheng, Xueqi",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.364/",
pages = "7621--7642",
ISBN = "979-8-89176-462-0",
abstract = "Text watermarking enables provenance tracing of LLM-generated text by embedding detectable signals, while multi-bit watermarking further encodes payloads such as user identities or model versions. Scaling multi-bit watermarking to large payloads is challenging as exhaustive decoding grows exponentially with payload length. Existing methods commonly address this by segmenting the payload and decoding each segment independently, but the impact of this design on decoding accuracy remains insufficiently understood. In this paper, we revisit segmentation in multi-bit watermarking and reveal a drawback of the commonly adopted paradigm: segmentation splits the available token evidence among segments, leaving fewer observations for each local decoding decision and thereby reducing decoding reliability. We theoretically characterize this accuracy{--}efficiency trade-off and empirically validate the claim across diverse settings. Additionally, we introduce a distribution-balanced detector to mitigate message-dependent bias. Experiments across multiple datasets and models show that our method achieves improved accuracy and robustness under editing attacks compared to existing methods. Our code is available at https://github.com/YouduLiu/SDBMW"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="huang-etal-2026-revisiting">
<titleInfo>
<title>Revisiting Segmentation in Multi-bit LLM Watermarking</title>
</titleInfo>
<name type="personal">
<namePart type="given">Beining</namePart>
<namePart type="family">Huang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Youdu</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fei</namePart>
<namePart type="family">Sun</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Du</namePart>
<namePart type="family">Su</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shuangjie</namePart>
<namePart type="family">Fu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huawei</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xueqi</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Text watermarking enables provenance tracing of LLM-generated text by embedding detectable signals, while multi-bit watermarking further encodes payloads such as user identities or model versions. Scaling multi-bit watermarking to large payloads is challenging as exhaustive decoding grows exponentially with payload length. Existing methods commonly address this by segmenting the payload and decoding each segment independently, but the impact of this design on decoding accuracy remains insufficiently understood. In this paper, we revisit segmentation in multi-bit watermarking and reveal a drawback of the commonly adopted paradigm: segmentation splits the available token evidence among segments, leaving fewer observations for each local decoding decision and thereby reducing decoding reliability. We theoretically characterize this accuracy–efficiency trade-off and empirically validate the claim across diverse settings. Additionally, we introduce a distribution-balanced detector to mitigate message-dependent bias. Experiments across multiple datasets and models show that our method achieves improved accuracy and robustness under editing attacks compared to existing methods. Our code is available at https://github.com/YouduLiu/SDBMW</abstract>
<identifier type="citekey">huang-etal-2026-revisiting</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.364/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>7621</start>
<end>7642</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Revisiting Segmentation in Multi-bit LLM Watermarking
%A Huang, Beining
%A Liu, Youdu
%A Sun, Fei
%A Su, Du
%A Fu, Shuangjie
%A Shen, Huawei
%A Cheng, Xueqi
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F huang-etal-2026-revisiting
%X Text watermarking enables provenance tracing of LLM-generated text by embedding detectable signals, while multi-bit watermarking further encodes payloads such as user identities or model versions. Scaling multi-bit watermarking to large payloads is challenging as exhaustive decoding grows exponentially with payload length. Existing methods commonly address this by segmenting the payload and decoding each segment independently, but the impact of this design on decoding accuracy remains insufficiently understood. In this paper, we revisit segmentation in multi-bit watermarking and reveal a drawback of the commonly adopted paradigm: segmentation splits the available token evidence among segments, leaving fewer observations for each local decoding decision and thereby reducing decoding reliability. We theoretically characterize this accuracy–efficiency trade-off and empirically validate the claim across diverse settings. Additionally, we introduce a distribution-balanced detector to mitigate message-dependent bias. Experiments across multiple datasets and models show that our method achieves improved accuracy and robustness under editing attacks compared to existing methods. Our code is available at https://github.com/YouduLiu/SDBMW
%U https://aclanthology.org/2026.findings-emnlp.364/
%P 7621-7642
Markdown (Informal)
[Revisiting Segmentation in Multi-bit LLM Watermarking](https://aclanthology.org/2026.findings-emnlp.364/) (Huang et al., Findings 2026)
ACL
- Beining Huang, Youdu Liu, Fei Sun, Du Su, Shuangjie Fu, Huawei Shen, and Xueqi Cheng. 2026. Revisiting Segmentation in Multi-bit LLM Watermarking. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 7621–7642, Budapest, Hungary. Association for Computational Linguistics.