<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="rajesh-etal-2026-convo">
    <titleInfo>
        <title>No Convo! The Illusion of Reliability in Online Toxicity Detection</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Rachita</namePart>
        <namePart type="family">Rajesh</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Christina</namePart>
        <namePart type="given">A</namePart>
        <namePart type="family">Chance</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ashima</namePart>
        <namePart type="family">Suvarna</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Kai-Wei</namePart>
        <namePart type="family">Chang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Saadia</namePart>
        <namePart type="family">Gabriel</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>While AI toxicity detection models are widely used, their (un)reliability under realistic linguistic variation remains poorly understood. We provide a comprehensive study of how contextual change affects prediction reliability. We evaluate eight state-of-the-art toxicity detection models across three dimensions of linguistic variation: conversational context, conversational depth, and surface form transformations. Our results on three toxicity benchmarks show that conversational context often introduces substantial prediction instability, with some systems exhibiting prediction flip rates of up to 40%, while increasing conversational depth fails to produce consistent performance gains. Under paraphrasing transformations designed to preserve the underlying meaning and harmful intent, models exhibit substantial degradation and prediction flip rates of up to 57%. Overall, our findings show that strong benchmark performance does not necessarily translate to robust moderation behavior under realistic linguistic variation, highlighting the need for robustness-oriented model development and evaluation frameworks that better capture deployment conditions.</abstract>
    <identifier type="citekey">rajesh-etal-2026-convo</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-aacl.204/</url>
    </location>
    <part>
        <date>2026-11</date>
        <extent unit="page">
            <start>4050</start>
            <end>4072</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T No Convo! The Illusion of Reliability in Online Toxicity Detection
%A Rajesh, Rachita
%A Chance, Christina A.
%A Suvarna, Ashima
%A Chang, Kai-Wei
%A Gabriel, Saadia
%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 rajesh-etal-2026-convo
%X While AI toxicity detection models are widely used, their (un)reliability under realistic linguistic variation remains poorly understood. We provide a comprehensive study of how contextual change affects prediction reliability. We evaluate eight state-of-the-art toxicity detection models across three dimensions of linguistic variation: conversational context, conversational depth, and surface form transformations. Our results on three toxicity benchmarks show that conversational context often introduces substantial prediction instability, with some systems exhibiting prediction flip rates of up to 40%, while increasing conversational depth fails to produce consistent performance gains. Under paraphrasing transformations designed to preserve the underlying meaning and harmful intent, models exhibit substantial degradation and prediction flip rates of up to 57%. Overall, our findings show that strong benchmark performance does not necessarily translate to robust moderation behavior under realistic linguistic variation, highlighting the need for robustness-oriented model development and evaluation frameworks that better capture deployment conditions.
%U https://aclanthology.org/2026.findings-aacl.204/
%P 4050-4072

Markdown (Informal)

[No Convo! The Illusion of Reliability in Online Toxicity Detection](https://aclanthology.org/2026.findings-aacl.204/) (Rajesh et al., Findings 2026)

• No Convo! The Illusion of Reliability in Online Toxicity Detection (Rajesh et al., Findings 2026)

ACL

• Rachita Rajesh, Christina A Chance, Ashima Suvarna, Kai-Wei Chang, and Saadia Gabriel. 2026. No Convo! The Illusion of Reliability in Online Toxicity Detection. In Findings of the Association for Computational Linguistics: IJCNLP-AACL 2026, pages 4050–4072, Hengqin, China. Association for Computational Linguistics.