AlbanianLLMSafety: A Safety Evaluation Dataset for Large Language Models in Albanian

Wajdi Zaghouani, Kholoud Khalil Aldous, Isra Fejzullaj


Abstract
Safety evaluation of Large Language Models (LLMs) has largely focused on high-resource languages, leaving low-resource languages critically underserved. We present AlbanianLLMSafety, the first publicly available safety evaluation dataset for LLMs in Albanian, a linguistically distinct low-resource language with approximately 7.5 million speakers across Albania, Kosovo, North Macedonia, and the diaspora. The dataset contains 2,951 prompts spanning 11 safety categories, including self-harm, violence, racist content, child exploitation, and radicalization, with an average of 268 prompts per category. Each prompt is provided in Albanian with an English reference translation and a detailed category label. This resource addresses a significant gap in safety evaluation infrastructure for low-resource languages and provides an essential benchmark for developing safer, more inclusive LLMs. The dataset will be provided upon request to support safety evaluation, fine-tuning, red-teaming, and guardrail development for Albanian-speaking communities.
Anthology ID:
2026.sigul-1.7
Volume:
Proceedings of the SIGUL 2026 Joint Workshop with ELE, EURALI, and DCLRL: Towards Inclusivity and Equality: Language Resources and Technologies for Under-Resourced and Endangered Languages
Month:
May
Year:
2026
Address:
Palma, Mallorca, Spain
Editors:
Atul Kr. Ojha, Sakriani Sakti, Claudia Soria, Maite Melero, John P. McCrae, Constantine Lignos, Chao-Hong Liu, German Rigau Claramunt, Georg Rehm
Venues:
SIGUL | EURALI | DCLRL | WS
SIG:
Publisher:
ELRA Language Resources Association (ELRA)
Note:
Pages:
66–73
Language:
External URL:
https://lrec.elra.info/lrec2026-ws-sigul-07
DOI:
10.63317/42i5uiy6yyqq
Bibkey:
Cite (ACL):
Wajdi Zaghouani, Kholoud Khalil Aldous, and Isra Fejzullaj. 2026. AlbanianLLMSafety: A Safety Evaluation Dataset for Large Language Models in Albanian. In Proceedings of the SIGUL 2026 Joint Workshop with ELE, EURALI, and DCLRL: Towards Inclusivity and Equality: Language Resources and Technologies for Under-Resourced and Endangered Languages, pages 66–73, Palma, Mallorca, Spain. ELRA Language Resources Association (ELRA).
Cite (Informal):
AlbanianLLMSafety: A Safety Evaluation Dataset for Large Language Models in Albanian (Zaghouani et al., SIGUL-EURALI-DCLRL 2026)
Copy Citation: