<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ceppi-sanchez-2026-flip">
    <titleInfo>
        <title>Flip, Don’t Shuffle: Watermarking LLMs at the Speed of Inference</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Simone</namePart>
        <namePart type="family">Ceppi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ignacio</namePart>
        <namePart type="family">Sanchez</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>We introduce Stateless Bernoulli Watermarking (SBW), a new statistical watermark for Large Language Models that determines green list membership through independent per-token Bernoulli trials. Unlike KGW’s vocabulary permutation or SynthID’s multi-layer tournament, SBW requires only a single comparison per token against a counter-based random number generator, reducing membership complexity to O(1) and enabling single-kernel execution with zero intermediate allocations. We prove that this formulation preserves the same detection guarantees as fixed-size green lists: the z-score test remains \mathcalN(0,1) under the null. The stateless architecture enables capabilities unavailable to existing methods: full-vocabulary self-salt watermarking (over 6000\times faster than KGW’s self-salt and 2\times faster than SynthID despite biasing the entire vocabulary with candidate-dependent seeding) and architectural compatibility with distributed inference. In end-to-end generation benchmarks, SBW adds less than 1% overhead at all batch sizes. We additionally identify hash function design as a previously unexplored axis for watermark quality, showing that a GPU-native Jenkins hash improves null calibration by 1.8\times while producing more diverse text. Experiments across two seeding schemes and eight (γ, δ) configurations confirm statistical equivalence with ROC-AUC differences below 0.01.</abstract>
    <identifier type="citekey">ceppi-sanchez-2026-flip</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.2124/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>48312</start>
            <end>48328</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Flip, Don’t Shuffle: Watermarking LLMs at the Speed of Inference
%A Ceppi, Simone
%A Sanchez, Ignacio
%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 ceppi-sanchez-2026-flip
%X We introduce Stateless Bernoulli Watermarking (SBW), a new statistical watermark for Large Language Models that determines green list membership through independent per-token Bernoulli trials. Unlike KGW’s vocabulary permutation or SynthID’s multi-layer tournament, SBW requires only a single comparison per token against a counter-based random number generator, reducing membership complexity to O(1) and enabling single-kernel execution with zero intermediate allocations. We prove that this formulation preserves the same detection guarantees as fixed-size green lists: the z-score test remains \mathcalN(0,1) under the null. The stateless architecture enables capabilities unavailable to existing methods: full-vocabulary self-salt watermarking (over 6000\times faster than KGW’s self-salt and 2\times faster than SynthID despite biasing the entire vocabulary with candidate-dependent seeding) and architectural compatibility with distributed inference. In end-to-end generation benchmarks, SBW adds less than 1% overhead at all batch sizes. We additionally identify hash function design as a previously unexplored axis for watermark quality, showing that a GPU-native Jenkins hash improves null calibration by 1.8\times while producing more diverse text. Experiments across two seeding schemes and eight (γ, δ) configurations confirm statistical equivalence with ROC-AUC differences below 0.01.
%U https://aclanthology.org/2026.emnlp-main.2124/
%P 48312-48328

Markdown (Informal)

[Flip, Don’t Shuffle: Watermarking LLMs at the Speed of Inference](https://aclanthology.org/2026.emnlp-main.2124/) (Ceppi & Sanchez, EMNLP 2026)

• Flip, Don’t Shuffle: Watermarking LLMs at the Speed of Inference (Ceppi & Sanchez, EMNLP 2026)

ACL

• Simone Ceppi and Ignacio Sanchez. 2026. Flip, Don’t Shuffle: Watermarking LLMs at the Speed of Inference. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 48312–48328, Budapest, Hungary. Association for Computational Linguistics.