@inproceedings{leng-etal-2026-rw2unsafe,
title = "{RW}2{U}nsafe: Towards More Accurate Real-World Unsafe Generation Evaluation for {T}2{I} Models",
author = "Leng, Ye and
Chu, Junjie and
Li, Mingjie and
Backes, Michael and
Shen, Yun and
Zhang, Yang",
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.1359/",
pages = "28860--28873",
ISBN = "979-8-89176-462-0",
abstract = "In-the-real-world datasets provide a crucial reflection of real-world adversarial distributions and hold significant practical importance for evaluating the safety of T2I models. However, existing in-the-real-world unsafe benchmarks are not tailored for T2I generation: many prompts lack visual specificity or concrete scenarios, producing distorted and low-fidelity images that are misclassified as safe by automated detectors. This acts as a confounder of successful safety alignment and compromises the accuracy of safety measurement. To address this, we propose {RW2Unsafe}, an automated three-module framework that reformulates source in-the-real-world posts into visually descriptive T2I prompts while preserving the original harmful intent. We use it to construct {rw{\_}t2i{\_}4chan}, a benchmark of 2,000 high-quality in-the-real-world unsafe T2I prompts. Auditing five advanced T2I models, we find that prior approaches relying on source prompts systematically underestimate safety risks: on Bagel, for instance, the mean unsafe score under source prompts is under one-third of that under rewritten prompts. We hope the proposed framework and dataset could contribute to more accurate safety measurement and safer T2I models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="leng-etal-2026-rw2unsafe">
<titleInfo>
<title>RW2Unsafe: Towards More Accurate Real-World Unsafe Generation Evaluation for T2I Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ye</namePart>
<namePart type="family">Leng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junjie</namePart>
<namePart type="family">Chu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingjie</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Michael</namePart>
<namePart type="family">Backes</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun</namePart>
<namePart type="family">Shen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yang</namePart>
<namePart type="family">Zhang</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>In-the-real-world datasets provide a crucial reflection of real-world adversarial distributions and hold significant practical importance for evaluating the safety of T2I models. However, existing in-the-real-world unsafe benchmarks are not tailored for T2I generation: many prompts lack visual specificity or concrete scenarios, producing distorted and low-fidelity images that are misclassified as safe by automated detectors. This acts as a confounder of successful safety alignment and compromises the accuracy of safety measurement. To address this, we propose RW2Unsafe, an automated three-module framework that reformulates source in-the-real-world posts into visually descriptive T2I prompts while preserving the original harmful intent. We use it to construct rw_t2i_4chan, a benchmark of 2,000 high-quality in-the-real-world unsafe T2I prompts. Auditing five advanced T2I models, we find that prior approaches relying on source prompts systematically underestimate safety risks: on Bagel, for instance, the mean unsafe score under source prompts is under one-third of that under rewritten prompts. We hope the proposed framework and dataset could contribute to more accurate safety measurement and safer T2I models.</abstract>
<identifier type="citekey">leng-etal-2026-rw2unsafe</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1359/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>28860</start>
<end>28873</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T RW2Unsafe: Towards More Accurate Real-World Unsafe Generation Evaluation for T2I Models
%A Leng, Ye
%A Chu, Junjie
%A Li, Mingjie
%A Backes, Michael
%A Shen, Yun
%A Zhang, Yang
%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 leng-etal-2026-rw2unsafe
%X In-the-real-world datasets provide a crucial reflection of real-world adversarial distributions and hold significant practical importance for evaluating the safety of T2I models. However, existing in-the-real-world unsafe benchmarks are not tailored for T2I generation: many prompts lack visual specificity or concrete scenarios, producing distorted and low-fidelity images that are misclassified as safe by automated detectors. This acts as a confounder of successful safety alignment and compromises the accuracy of safety measurement. To address this, we propose RW2Unsafe, an automated three-module framework that reformulates source in-the-real-world posts into visually descriptive T2I prompts while preserving the original harmful intent. We use it to construct rw_t2i_4chan, a benchmark of 2,000 high-quality in-the-real-world unsafe T2I prompts. Auditing five advanced T2I models, we find that prior approaches relying on source prompts systematically underestimate safety risks: on Bagel, for instance, the mean unsafe score under source prompts is under one-third of that under rewritten prompts. We hope the proposed framework and dataset could contribute to more accurate safety measurement and safer T2I models.
%U https://aclanthology.org/2026.findings-emnlp.1359/
%P 28860-28873
Markdown (Informal)
[RW2Unsafe: Towards More Accurate Real-World Unsafe Generation Evaluation for T2I Models](https://aclanthology.org/2026.findings-emnlp.1359/) (Leng et al., Findings 2026)
ACL