@inproceedings{zhou-etal-2025-kill,
title = "Kill two birds with one stone: generalized and robust {AI}-generated text detection via dynamic perturbations",
author = "Zhou, Yinghan and
Wen, Juan and
Peng, Wanli and
Yiming, Xue and
Zhang, ZiWei and
Zhengxian, Wu",
editor = "Chiruzzo, Luis and
Ritter, Alan and
Wang, Lu",
booktitle = "Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)",
month = apr,
year = "2025",
address = "Albuquerque, New Mexico",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.naacl-long.446/",
doi = "10.18653/v1/2025.naacl-long.446",
pages = "8864--8875",
ISBN = "979-8-89176-189-6",
abstract = "The growing popularity of large language models has raised concerns regarding the potential to misuse AI-generated text (AIGT). It becomes increasingly critical to establish an excellent AIGT detection method with high generalization and robustness.While, existing methods either focus on model generalization or concentrate on robustness.The unified mechanism, to simultaneously address the challenges of generalization and robustness, is less explored. In this paper, we first empirically reveal an intrinsic mechanism for model generalization and robustness of AIGT detection task.Then, we proposed a novel AIGT detection method (DP-Net) via dynamic perturbations introduced by a reinforcement learning with elaborated reward and action.Experimentally, extensive results show that the proposed DP-Net significantly outperforms some state-of-the-art AIGT detection methods for generalization capacity in three cross-domain scenarios.Meanwhile, the DP-Net achieves best robustness under two text adversarial attacks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zhou-etal-2025-kill">
<titleInfo>
<title>Kill two birds with one stone: generalized and robust AI-generated text detection via dynamic perturbations</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yinghan</namePart>
<namePart type="family">Zhou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juan</namePart>
<namePart type="family">Wen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wanli</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xue</namePart>
<namePart type="family">Yiming</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">ZiWei</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wu</namePart>
<namePart type="family">Zhengxian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-04</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Luis</namePart>
<namePart type="family">Chiruzzo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Alan</namePart>
<namePart type="family">Ritter</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lu</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Albuquerque, New Mexico</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-189-6</identifier>
</relatedItem>
<abstract>The growing popularity of large language models has raised concerns regarding the potential to misuse AI-generated text (AIGT). It becomes increasingly critical to establish an excellent AIGT detection method with high generalization and robustness.While, existing methods either focus on model generalization or concentrate on robustness.The unified mechanism, to simultaneously address the challenges of generalization and robustness, is less explored. In this paper, we first empirically reveal an intrinsic mechanism for model generalization and robustness of AIGT detection task.Then, we proposed a novel AIGT detection method (DP-Net) via dynamic perturbations introduced by a reinforcement learning with elaborated reward and action.Experimentally, extensive results show that the proposed DP-Net significantly outperforms some state-of-the-art AIGT detection methods for generalization capacity in three cross-domain scenarios.Meanwhile, the DP-Net achieves best robustness under two text adversarial attacks.</abstract>
<identifier type="citekey">zhou-etal-2025-kill</identifier>
<identifier type="doi">10.18653/v1/2025.naacl-long.446</identifier>
<location>
<url>https://aclanthology.org/2025.naacl-long.446/</url>
</location>
<part>
<date>2025-04</date>
<extent unit="page">
<start>8864</start>
<end>8875</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Kill two birds with one stone: generalized and robust AI-generated text detection via dynamic perturbations
%A Zhou, Yinghan
%A Wen, Juan
%A Peng, Wanli
%A Yiming, Xue
%A Zhang, ZiWei
%A Zhengxian, Wu
%Y Chiruzzo, Luis
%Y Ritter, Alan
%Y Wang, Lu
%S Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
%D 2025
%8 April
%I Association for Computational Linguistics
%C Albuquerque, New Mexico
%@ 979-8-89176-189-6
%F zhou-etal-2025-kill
%X The growing popularity of large language models has raised concerns regarding the potential to misuse AI-generated text (AIGT). It becomes increasingly critical to establish an excellent AIGT detection method with high generalization and robustness.While, existing methods either focus on model generalization or concentrate on robustness.The unified mechanism, to simultaneously address the challenges of generalization and robustness, is less explored. In this paper, we first empirically reveal an intrinsic mechanism for model generalization and robustness of AIGT detection task.Then, we proposed a novel AIGT detection method (DP-Net) via dynamic perturbations introduced by a reinforcement learning with elaborated reward and action.Experimentally, extensive results show that the proposed DP-Net significantly outperforms some state-of-the-art AIGT detection methods for generalization capacity in three cross-domain scenarios.Meanwhile, the DP-Net achieves best robustness under two text adversarial attacks.
%R 10.18653/v1/2025.naacl-long.446
%U https://aclanthology.org/2025.naacl-long.446/
%U https://doi.org/10.18653/v1/2025.naacl-long.446
%P 8864-8875
Markdown (Informal)
[Kill two birds with one stone: generalized and robust AI-generated text detection via dynamic perturbations](https://aclanthology.org/2025.naacl-long.446/) (Zhou et al., NAACL 2025)
ACL