<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="han-etal-2026-implicit-unwatermarking">
    <titleInfo>
        <title>Implicit Unwatermarking in Pseudo-Reference Reranking for LLM Watermarking</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Sungwoo</namePart>
        <namePart type="family">Han</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jingun</namePart>
        <namePart type="family">Kwon</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Hidetaka</namePart>
        <namePart type="family">Kamigaito</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Manabu</namePart>
        <namePart type="family">Okumura</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>Text watermarking for large language models (LLMs) requires balancing output quality and detection reliability. We study a practically relevant reranking setting in which multiple watermarked outputs are compared against unwatermarked pseudo-references. We show that quality-only reranking can weaken watermark signals by favoring candidates closer to unwatermarked references, effectively acting as implicit unwatermarking. We propose constrained Minimum Bayes Risk (MBR) selection, which treats detection as a constraint rather than a weighted objective: it filters out weakly watermarked candidates and then selects the highest-quality remaining output. Experiments across watermarking methods, datasets, and models show that constrained selection substantially improves detectability over quality-only reranking while remaining competitive in task quality. It also achieves a more favorable matched-detection tradeoff than normalized weighted MBR.</abstract>
    <identifier type="citekey">han-etal-2026-implicit-unwatermarking</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.2206/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>50175</start>
            <end>50187</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Implicit Unwatermarking in Pseudo-Reference Reranking for LLM Watermarking
%A Han, Sungwoo
%A Kwon, Jingun
%A Kamigaito, Hidetaka
%A Okumura, Manabu
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F han-etal-2026-implicit-unwatermarking
%X Text watermarking for large language models (LLMs) requires balancing output quality and detection reliability. We study a practically relevant reranking setting in which multiple watermarked outputs are compared against unwatermarked pseudo-references. We show that quality-only reranking can weaken watermark signals by favoring candidates closer to unwatermarked references, effectively acting as implicit unwatermarking. We propose constrained Minimum Bayes Risk (MBR) selection, which treats detection as a constraint rather than a weighted objective: it filters out weakly watermarked candidates and then selects the highest-quality remaining output. Experiments across watermarking methods, datasets, and models show that constrained selection substantially improves detectability over quality-only reranking while remaining competitive in task quality. It also achieves a more favorable matched-detection tradeoff than normalized weighted MBR.
%U https://aclanthology.org/2026.emnlp-main.2206/
%P 50175-50187

Markdown (Informal)

[Implicit Unwatermarking in Pseudo-Reference Reranking for LLM Watermarking](https://aclanthology.org/2026.emnlp-main.2206/) (Han et al., EMNLP 2026)

• Implicit Unwatermarking in Pseudo-Reference Reranking for LLM Watermarking (Han et al., EMNLP 2026)

ACL

• Sungwoo Han, Jingun Kwon, Hidetaka Kamigaito, and Manabu Okumura. 2026. Implicit Unwatermarking in Pseudo-Reference Reranking for LLM Watermarking. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 50175–50187, Budapest, Hungary. Association for Computational Linguistics.