@inproceedings{hu-etal-2026-coat,
title = "{COAT}: An Orthogonal Benchmark for Diagnosing {C}hinese Adversarial Toxicity Detection Failures",
author = "Hu, Jiaqi and
Liu, Mingshuai and
Wu, Haobo and
Li, Xin",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {IJCNLP}-{AACL} 2026",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-aacl.173/",
pages = "3453--3465",
ISBN = "979-8-89176-444-6",
abstract = "Character-level perturbations are widely used to evade Chinese toxicity detection, yet existing benchmarks either cover only a limited range of perturbations or expand perturbation coverage without balanced construction across toxic and non-toxic samples, leaving these two requirements unmet within a single benchmark. We introduce COAT (Chinese Orthogonal Adversarial Toxicity), a 41K-sample benchmark spanning eight perturbation types that enforces perturbation orthogonality: all perturbations are applied with equal probability to both toxic and non-toxic texts, ensuring P(toxic | perturbed) = 0.5. As a core diagnostic layer, we annotate Natural Language Variant Ambiguity (NLVA): 21.6{\%} of clean samples naturally contain surface features overlapping with perturbation artifacts, enabling controlled analysis of model sensitivity to perturbation-like features and classification difficulty. Evaluating five encoder classifiers, three proprietary LLMs, and Llama3-8B with chain-of-thought reasoning, we find that (1) models remain systematically sensitive to perturbation-like features even under balanced training, and (2) across Llama3-8B variants, overcorrection varies systematically with the linguistic ambiguity introduced by each perturbation type. Step-level diagnosis localizes the shared bottleneck to semantic restoration, with targeted preference optimization yielding reliable gains on the most ambiguous types."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="hu-etal-2026-coat">
<titleInfo>
<title>COAT: An Orthogonal Benchmark for Diagnosing Chinese Adversarial Toxicity Detection Failures</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jiaqi</namePart>
<namePart type="family">Hu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingshuai</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Haobo</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xin</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-444-6</identifier>
</relatedItem>
<abstract>Character-level perturbations are widely used to evade Chinese toxicity detection, yet existing benchmarks either cover only a limited range of perturbations or expand perturbation coverage without balanced construction across toxic and non-toxic samples, leaving these two requirements unmet within a single benchmark. We introduce COAT (Chinese Orthogonal Adversarial Toxicity), a 41K-sample benchmark spanning eight perturbation types that enforces perturbation orthogonality: all perturbations are applied with equal probability to both toxic and non-toxic texts, ensuring P(toxic | perturbed) = 0.5. As a core diagnostic layer, we annotate Natural Language Variant Ambiguity (NLVA): 21.6% of clean samples naturally contain surface features overlapping with perturbation artifacts, enabling controlled analysis of model sensitivity to perturbation-like features and classification difficulty. Evaluating five encoder classifiers, three proprietary LLMs, and Llama3-8B with chain-of-thought reasoning, we find that (1) models remain systematically sensitive to perturbation-like features even under balanced training, and (2) across Llama3-8B variants, overcorrection varies systematically with the linguistic ambiguity introduced by each perturbation type. Step-level diagnosis localizes the shared bottleneck to semantic restoration, with targeted preference optimization yielding reliable gains on the most ambiguous types.</abstract>
<identifier type="citekey">hu-etal-2026-coat</identifier>
<location>
<url>https://aclanthology.org/2026.findings-aacl.173/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>3453</start>
<end>3465</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T COAT: An Orthogonal Benchmark for Diagnosing Chinese Adversarial Toxicity Detection Failures
%A Hu, Jiaqi
%A Liu, Mingshuai
%A Wu, Haobo
%A Li, Xin
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-444-6
%F hu-etal-2026-coat
%X Character-level perturbations are widely used to evade Chinese toxicity detection, yet existing benchmarks either cover only a limited range of perturbations or expand perturbation coverage without balanced construction across toxic and non-toxic samples, leaving these two requirements unmet within a single benchmark. We introduce COAT (Chinese Orthogonal Adversarial Toxicity), a 41K-sample benchmark spanning eight perturbation types that enforces perturbation orthogonality: all perturbations are applied with equal probability to both toxic and non-toxic texts, ensuring P(toxic | perturbed) = 0.5. As a core diagnostic layer, we annotate Natural Language Variant Ambiguity (NLVA): 21.6% of clean samples naturally contain surface features overlapping with perturbation artifacts, enabling controlled analysis of model sensitivity to perturbation-like features and classification difficulty. Evaluating five encoder classifiers, three proprietary LLMs, and Llama3-8B with chain-of-thought reasoning, we find that (1) models remain systematically sensitive to perturbation-like features even under balanced training, and (2) across Llama3-8B variants, overcorrection varies systematically with the linguistic ambiguity introduced by each perturbation type. Step-level diagnosis localizes the shared bottleneck to semantic restoration, with targeted preference optimization yielding reliable gains on the most ambiguous types.
%U https://aclanthology.org/2026.findings-aacl.173/
%P 3453-3465
Markdown (Informal)
[COAT: An Orthogonal Benchmark for Diagnosing Chinese Adversarial Toxicity Detection Failures](https://aclanthology.org/2026.findings-aacl.173/) (Hu et al., Findings 2026)
ACL