RefusEU dataset from "Multilingual Refusal Alignment for Safer Large Language Models" by A. Krasnodębska, W. Kusa, A. Lipani (ACL Findings 2026).