“Evaluating Generative AI Safety on Real-World Harm: A Focused Classification Study Using the RealHarm Dataset”. 2026. AI and Wisdom 1 (2). https://www.aiandwisdom.org/index.php/aiandwisdom/article/view/14.