@inproceedings{ma-etal-2026-disgar,
title = "{D}is{GAR}: Machine Generated Text Detection via Disentangled Generative Artifact Representation",
author = "Ma, Shuhua and
Li, Bohao and
Zhang, Lingcui and
He, Yanru and
Niu, Ben and
Li, Fenghua",
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.1997/",
doi = "10.18653/2026.findings-emnlp.1997",
pages = "42569--42583",
ISBN = "979-8-89176-462-0",
abstract = "The widespread misuse of large language models (LLMs) has made machine-generated text detection (MGTD) a critical task. Existing detection methods are often vulnerable to stylistic noise, which limits their generalization in real-world scenarios and robustness on short texts. Furthermore, many methods generate multiple auxiliary texts through LLM-based rewriting or similar strategies, substantially increasing detection latency and limiting their practicality. To address these challenges, we propose the **DisGAR** detection framework, which strips stylistic noise from the input text to mitigate its impact on detection performance, thereby improving cross-domain and cross-model robustness. Specifically, we first fine-tune the attention layers of the proxy model to capture generation-specific patterns. Then we further introduce a global artifact aggregation mechanism and a contrastive learning strategy based on rewritten texts, enabling the separation of stylistic artifacts from the underlying generation patterns. Experimental results demonstrate that DisGAR outperforms existing state-of-the-art methods in unseen-model and unseen-domain detection. Moreover, DisGAR achieves detection efficiency comparable to Fast-DetectGPT, successfully balancing detection performance and efficiency. Source code is available at https://github.com/m-Clown/DisGAR.git"
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ma-etal-2026-disgar">
<titleInfo>
<title>DisGAR: Machine Generated Text Detection via Disentangled Generative Artifact Representation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shuhua</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bohao</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lingcui</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yanru</namePart>
<namePart type="family">He</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ben</namePart>
<namePart type="family">Niu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fenghua</namePart>
<namePart type="family">Li</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>The widespread misuse of large language models (LLMs) has made machine-generated text detection (MGTD) a critical task. Existing detection methods are often vulnerable to stylistic noise, which limits their generalization in real-world scenarios and robustness on short texts. Furthermore, many methods generate multiple auxiliary texts through LLM-based rewriting or similar strategies, substantially increasing detection latency and limiting their practicality. To address these challenges, we propose the **DisGAR** detection framework, which strips stylistic noise from the input text to mitigate its impact on detection performance, thereby improving cross-domain and cross-model robustness. Specifically, we first fine-tune the attention layers of the proxy model to capture generation-specific patterns. Then we further introduce a global artifact aggregation mechanism and a contrastive learning strategy based on rewritten texts, enabling the separation of stylistic artifacts from the underlying generation patterns. Experimental results demonstrate that DisGAR outperforms existing state-of-the-art methods in unseen-model and unseen-domain detection. Moreover, DisGAR achieves detection efficiency comparable to Fast-DetectGPT, successfully balancing detection performance and efficiency. Source code is available at https://github.com/m-Clown/DisGAR.git</abstract>
<identifier type="citekey">ma-etal-2026-disgar</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1997</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1997/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>42569</start>
<end>42583</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T DisGAR: Machine Generated Text Detection via Disentangled Generative Artifact Representation
%A Ma, Shuhua
%A Li, Bohao
%A Zhang, Lingcui
%A He, Yanru
%A Niu, Ben
%A Li, Fenghua
%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 ma-etal-2026-disgar
%X The widespread misuse of large language models (LLMs) has made machine-generated text detection (MGTD) a critical task. Existing detection methods are often vulnerable to stylistic noise, which limits their generalization in real-world scenarios and robustness on short texts. Furthermore, many methods generate multiple auxiliary texts through LLM-based rewriting or similar strategies, substantially increasing detection latency and limiting their practicality. To address these challenges, we propose the **DisGAR** detection framework, which strips stylistic noise from the input text to mitigate its impact on detection performance, thereby improving cross-domain and cross-model robustness. Specifically, we first fine-tune the attention layers of the proxy model to capture generation-specific patterns. Then we further introduce a global artifact aggregation mechanism and a contrastive learning strategy based on rewritten texts, enabling the separation of stylistic artifacts from the underlying generation patterns. Experimental results demonstrate that DisGAR outperforms existing state-of-the-art methods in unseen-model and unseen-domain detection. Moreover, DisGAR achieves detection efficiency comparable to Fast-DetectGPT, successfully balancing detection performance and efficiency. Source code is available at https://github.com/m-Clown/DisGAR.git
%R 10.18653/2026.findings-emnlp.1997
%U https://aclanthology.org/2026.findings-emnlp.1997/
%U https://doi.org/10.18653/2026.findings-emnlp.1997
%P 42569-42583
Markdown (Informal)
[DisGAR: Machine Generated Text Detection via Disentangled Generative Artifact Representation](https://aclanthology.org/2026.findings-emnlp.1997/) (Ma et al., Findings 2026)
ACL