@inproceedings{akinode-etal-2026-tukabench,
title = "{T}uka{B}ench: A Culturally Grounded Jailbreak Benchmark for {A}frican Languages",
author = "Akinode, Victor Olalekan and
Li, Senyu and
Hamidouche, Wassim and
Zamir, Syed Waqas and
Becker-Reshef, Inbal and
Adelani, David Ifeoluwa",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.692/",
pages = "15907--15930",
ISBN = "979-8-89176-459-0",
abstract = "Safety evaluation of Large Language Models (LLMs) remains heavily English-centric, leaving Low-Resource Languages (LRLs), particularly African ones, critically underexplored. We introduce TUKABENCH, a jailbreak benchmark for seven African languages that extends JailbreakBench (JBB) beyond direct translation through four settings: human translation of JBB prompts, English adaptation to African contexts followed by human translation, human curated prompts validated through interaction with GPT-5.2 to filter out easily detectable patterns, and code-switched prompts combining English and African languages. These settings isolate the effects of language, cultural grounding, and prompt evasiveness on model safety. Across closed and open models, prompting in African languages reduces refusal relative to English, with culturally adapted prompts leading to least refusal. The evaluation also surfaces two structural limitations: model comprehension failures and reduced LLM-as-a-judge reliability in LRLs. To capture the first, we introduce Deflection alongside Refused and Jailbroken; to assess the second, we validate outputs with human annotations, showing that judge-human agreement drops in lower-resource languages and less commonly supported scripts."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="akinode-etal-2026-tukabench">
<titleInfo>
<title>TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages</title>
</titleInfo>
<name type="personal">
<namePart type="given">Victor</namePart>
<namePart type="given">Olalekan</namePart>
<namePart type="family">Akinode</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Senyu</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wassim</namePart>
<namePart type="family">Hamidouche</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Syed</namePart>
<namePart type="given">Waqas</namePart>
<namePart type="family">Zamir</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Inbal</namePart>
<namePart type="family">Becker-Reshef</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="given">Ifeoluwa</namePart>
<namePart type="family">Adelani</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>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Safety evaluation of Large Language Models (LLMs) remains heavily English-centric, leaving Low-Resource Languages (LRLs), particularly African ones, critically underexplored. We introduce TUKABENCH, a jailbreak benchmark for seven African languages that extends JailbreakBench (JBB) beyond direct translation through four settings: human translation of JBB prompts, English adaptation to African contexts followed by human translation, human curated prompts validated through interaction with GPT-5.2 to filter out easily detectable patterns, and code-switched prompts combining English and African languages. These settings isolate the effects of language, cultural grounding, and prompt evasiveness on model safety. Across closed and open models, prompting in African languages reduces refusal relative to English, with culturally adapted prompts leading to least refusal. The evaluation also surfaces two structural limitations: model comprehension failures and reduced LLM-as-a-judge reliability in LRLs. To capture the first, we introduce Deflection alongside Refused and Jailbroken; to assess the second, we validate outputs with human annotations, showing that judge-human agreement drops in lower-resource languages and less commonly supported scripts.</abstract>
<identifier type="citekey">akinode-etal-2026-tukabench</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.692/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>15907</start>
<end>15930</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages
%A Akinode, Victor Olalekan
%A Li, Senyu
%A Hamidouche, Wassim
%A Zamir, Syed Waqas
%A Becker-Reshef, Inbal
%A Adelani, David Ifeoluwa
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F akinode-etal-2026-tukabench
%X Safety evaluation of Large Language Models (LLMs) remains heavily English-centric, leaving Low-Resource Languages (LRLs), particularly African ones, critically underexplored. We introduce TUKABENCH, a jailbreak benchmark for seven African languages that extends JailbreakBench (JBB) beyond direct translation through four settings: human translation of JBB prompts, English adaptation to African contexts followed by human translation, human curated prompts validated through interaction with GPT-5.2 to filter out easily detectable patterns, and code-switched prompts combining English and African languages. These settings isolate the effects of language, cultural grounding, and prompt evasiveness on model safety. Across closed and open models, prompting in African languages reduces refusal relative to English, with culturally adapted prompts leading to least refusal. The evaluation also surfaces two structural limitations: model comprehension failures and reduced LLM-as-a-judge reliability in LRLs. To capture the first, we introduce Deflection alongside Refused and Jailbroken; to assess the second, we validate outputs with human annotations, showing that judge-human agreement drops in lower-resource languages and less commonly supported scripts.
%U https://aclanthology.org/2026.emnlp-main.692/
%P 15907-15930
Markdown (Informal)
[TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages](https://aclanthology.org/2026.emnlp-main.692/) (Akinode et al., EMNLP 2026)
ACL
- Victor Olalekan Akinode, Senyu Li, Wassim Hamidouche, Syed Waqas Zamir, Inbal Becker-Reshef, and David Ifeoluwa Adelani. 2026. TukaBench: A Culturally Grounded Jailbreak Benchmark for African Languages. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 15907–15930, Budapest, Hungary. Association for Computational Linguistics.