@inproceedings{zaghouani-etal-2026-arabicdialectsafety,
title = "{A}rabic{D}ialect{S}afety: A Dialect-Aware Benchmark for {A}rabic Content Safety Classification",
author = "Zaghouani, Wajdi and
Biswas, Md. Rafiul and
Aldous, Kholoud Khalil and
Bessghaier, Mabrouka",
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.1726/",
pages = "36924--36938",
ISBN = "979-8-89176-462-0",
abstract = "We present ArabicDialectSafety, a human-curated Arabic safety dataset of 25,071 prompts covering six Arabic varieties{---}Modern Standard Arabic, Syrian, Egyptian, Algerian, Palestinian, and Moroccan{---}annotated with dialect labels and seven fine-grained harm categories. We introduce a dual-task evaluation framework for binary safe/unsafe detection and granular harm classification across dialects. Benchmarking seven supervised and generative models, we find that fine-tuned MARBERTv2 achieves the strongest performance (0.95 binary and 0.90 granular Macro-F1), substantially outperforming prompted frontier LLMs, including Arabic-specialized models. Our analyses show that dialect conditioning is most effective when integrated at the representation level, while significant performance gaps remain for low-resource Maghrebi dialects. We further evaluate seven frontier LLMs as response generators on harmful dialectal Arabic prompts and observe unsafe generation rates below 5{\%} across models. We release the dataset and code upon acceptance to support future research on dialect-aware Arabic safety evaluation. Warning: this paper contains examples of harmful and potentially offensive content included solely for research purposes."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="zaghouani-etal-2026-arabicdialectsafety">
<titleInfo>
<title>ArabicDialectSafety: A Dialect-Aware Benchmark for Arabic Content Safety Classification</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wajdi</namePart>
<namePart type="family">Zaghouani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Md.</namePart>
<namePart type="given">Rafiul</namePart>
<namePart type="family">Biswas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kholoud</namePart>
<namePart type="given">Khalil</namePart>
<namePart type="family">Aldous</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mabrouka</namePart>
<namePart type="family">Bessghaier</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>We present ArabicDialectSafety, a human-curated Arabic safety dataset of 25,071 prompts covering six Arabic varieties—Modern Standard Arabic, Syrian, Egyptian, Algerian, Palestinian, and Moroccan—annotated with dialect labels and seven fine-grained harm categories. We introduce a dual-task evaluation framework for binary safe/unsafe detection and granular harm classification across dialects. Benchmarking seven supervised and generative models, we find that fine-tuned MARBERTv2 achieves the strongest performance (0.95 binary and 0.90 granular Macro-F1), substantially outperforming prompted frontier LLMs, including Arabic-specialized models. Our analyses show that dialect conditioning is most effective when integrated at the representation level, while significant performance gaps remain for low-resource Maghrebi dialects. We further evaluate seven frontier LLMs as response generators on harmful dialectal Arabic prompts and observe unsafe generation rates below 5% across models. We release the dataset and code upon acceptance to support future research on dialect-aware Arabic safety evaluation. Warning: this paper contains examples of harmful and potentially offensive content included solely for research purposes.</abstract>
<identifier type="citekey">zaghouani-etal-2026-arabicdialectsafety</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1726/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>36924</start>
<end>36938</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ArabicDialectSafety: A Dialect-Aware Benchmark for Arabic Content Safety Classification
%A Zaghouani, Wajdi
%A Biswas, Md. Rafiul
%A Aldous, Kholoud Khalil
%A Bessghaier, Mabrouka
%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 zaghouani-etal-2026-arabicdialectsafety
%X We present ArabicDialectSafety, a human-curated Arabic safety dataset of 25,071 prompts covering six Arabic varieties—Modern Standard Arabic, Syrian, Egyptian, Algerian, Palestinian, and Moroccan—annotated with dialect labels and seven fine-grained harm categories. We introduce a dual-task evaluation framework for binary safe/unsafe detection and granular harm classification across dialects. Benchmarking seven supervised and generative models, we find that fine-tuned MARBERTv2 achieves the strongest performance (0.95 binary and 0.90 granular Macro-F1), substantially outperforming prompted frontier LLMs, including Arabic-specialized models. Our analyses show that dialect conditioning is most effective when integrated at the representation level, while significant performance gaps remain for low-resource Maghrebi dialects. We further evaluate seven frontier LLMs as response generators on harmful dialectal Arabic prompts and observe unsafe generation rates below 5% across models. We release the dataset and code upon acceptance to support future research on dialect-aware Arabic safety evaluation. Warning: this paper contains examples of harmful and potentially offensive content included solely for research purposes.
%U https://aclanthology.org/2026.findings-emnlp.1726/
%P 36924-36938
Markdown (Informal)
[ArabicDialectSafety: A Dialect-Aware Benchmark for Arabic Content Safety Classification](https://aclanthology.org/2026.findings-emnlp.1726/) (Zaghouani et al., Findings 2026)
ACL