This research introduces Qorgau, a novel dataset designed specifically for evaluating LLM safety in bilingual environments, addressing a critical gap in security research that has primarily focused on English-only contexts.

• Examines language- and region-specific risks that emerge in Kazakh-Russian bilingual settings
• Reveals how safety findings in bilingual contexts can differ significantly from monolingual evaluations
• Provides a framework for culturally-relevant safety assessment in Central Asian contexts
• Highlights the importance of culturally-specific approaches to AI safety evaluation

This work is essential for security professionals as it demonstrates that responsible AI deployment requires evaluating safety across diverse linguistic and cultural contexts, not just in dominant languages.

Original paper: Qorgau: Evaluating LLM Safety in Kazakh-Russian Bilingual Contexts