PulseAugur
EN
LIVE 10:51:40

New Kazakh Prompt Dataset Reveals LLM Safety Gaps

By PulseAugur Editorial · [2 sources] ·

Researchers have developed KZ-SafetyPrompts, a new dataset designed to evaluate the safety of large language models (LLMs) in the Kazakh language. The dataset comprises 5,717 prompts across eleven risk categories, including violence, hate speech, and illegal activities, with native Kazakh phrasing and English translations. Initial testing with GPT-4o revealed a 28.2% refusal rate, highlighting specific safety vulnerabilities in Kazakh language processing that are not apparent in English-only evaluations. AI

IMPACT This dataset could improve LLM safety for underrepresented languages, highlighting specific vulnerabilities in Kazakh.

RANK_REASON The cluster contains an academic paper detailing a new dataset for LLM safety evaluation.

Read on arXiv cs.CL →

AI-generated summary · Google Gemini · from 2 sources. How we write summaries →

New Kazakh Prompt Dataset Reveals LLM Safety Gaps

COVERAGE [2]

  1. arXiv cs.CL TIER_1 English(EN) · Wajdi Zaghouani, Shimaa Amer Ibrahim, Aruzhan Muratbek, Olzhasbek Zhakenov, Adiya Akhmetzhanova ·

    KZ-SafetyPrompts: A Kazakh Safety Evaluation Prompt Dataset for Large Language Models

    arXiv:2605.26947v1 Announce Type: new Abstract: Kazakh is underrepresented in resources for evaluating the safety behavior of large language models. We present KZ-SafetyPrompts, a Kazakh prompt dataset for safety evaluation across eleven categories covering common risk areas such…

  2. arXiv cs.CL TIER_1 English(EN) · Adiya Akhmetzhanova ·

    KZ-SafetyPrompts: A Kazakh Safety Evaluation Prompt Dataset for Large Language Models

    Kazakh is underrepresented in resources for evaluating the safety behavior of large language models. We present KZ-SafetyPrompts, a Kazakh prompt dataset for safety evaluation across eleven categories covering common risk areas such as self-harm, violence, child exploitation, sex…

RELATED ENTITIES

RELATED TOPICS