As artificial intelligence (AI) becomes increasingly capable of assisting with health care tasks, a new study by researchers at the Icahn School of Medicine at Mount Sinai has found that adding a brief safety reminder reduced potentially harmful choices by AI models in clinical scenarios.