@inproceedings{schafer-etal-2026-robust,
title = "Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings",
author = {Sch{\"a}fer, Jonas and
Pilaszewicz, Cezary and
Wunder, Gerhard},
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.959/",
doi = "10.18653/2026.findings-emnlp.959",
pages = "20266--20287",
ISBN = "979-8-89176-462-0",
abstract = "This work presents Dual-Embedding Watermarking (DEW), a semantic watermarking scheme for large language models (LLMs) that leverages contextual and token-level embeddings to enhance robustness against paraphrasing and translation. DEW utilizes a signal-processing methodology, applying algebraic vector-space operations to token and context embeddings to derive a watermark signal that degrades gracefully under semantic shifts. The method obfuscates the watermark by projecting embedding vectors through pseudo-random matrices seeded with a secret key. Experimental results show that dual-embedding watermarking can offer state-of-the-art robustness, particularly against translation, while incurring relatively low computational overhead compared with other semantic schemes. At lower watermark strength, DEW also maintains competitive text quality, suggesting that dual-embedding signals provide a promising substrate for robust semantic watermarking."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="schafer-etal-2026-robust">
<titleInfo>
<title>Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jonas</namePart>
<namePart type="family">Schäfer</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cezary</namePart>
<namePart type="family">Pilaszewicz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Gerhard</namePart>
<namePart type="family">Wunder</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>This work presents Dual-Embedding Watermarking (DEW), a semantic watermarking scheme for large language models (LLMs) that leverages contextual and token-level embeddings to enhance robustness against paraphrasing and translation. DEW utilizes a signal-processing methodology, applying algebraic vector-space operations to token and context embeddings to derive a watermark signal that degrades gracefully under semantic shifts. The method obfuscates the watermark by projecting embedding vectors through pseudo-random matrices seeded with a secret key. Experimental results show that dual-embedding watermarking can offer state-of-the-art robustness, particularly against translation, while incurring relatively low computational overhead compared with other semantic schemes. At lower watermark strength, DEW also maintains competitive text quality, suggesting that dual-embedding signals provide a promising substrate for robust semantic watermarking.</abstract>
<identifier type="citekey">schafer-etal-2026-robust</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.959</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.959/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>20266</start>
<end>20287</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings
%A Schäfer, Jonas
%A Pilaszewicz, Cezary
%A Wunder, Gerhard
%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 schafer-etal-2026-robust
%X This work presents Dual-Embedding Watermarking (DEW), a semantic watermarking scheme for large language models (LLMs) that leverages contextual and token-level embeddings to enhance robustness against paraphrasing and translation. DEW utilizes a signal-processing methodology, applying algebraic vector-space operations to token and context embeddings to derive a watermark signal that degrades gracefully under semantic shifts. The method obfuscates the watermark by projecting embedding vectors through pseudo-random matrices seeded with a secret key. Experimental results show that dual-embedding watermarking can offer state-of-the-art robustness, particularly against translation, while incurring relatively low computational overhead compared with other semantic schemes. At lower watermark strength, DEW also maintains competitive text quality, suggesting that dual-embedding signals provide a promising substrate for robust semantic watermarking.
%R 10.18653/2026.findings-emnlp.959
%U https://aclanthology.org/2026.findings-emnlp.959/
%U https://doi.org/10.18653/2026.findings-emnlp.959
%P 20266-20287
Markdown (Informal)
[Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings](https://aclanthology.org/2026.findings-emnlp.959/) (Schäfer et al., Findings 2026)
ACL