Uncovering AI’s hidden risks: an empirical analysis of health-related AI incidents and their ethical implications

The rapid integration of artificial intelligence (AI) into healthcare has transformed clinical decision-making, digital therapeutics, and precision medicine, yet this technological revolution raises profound ethical questions about patient safety, transparency, and accountability in medical care. The implications of AI-related adverse events remain critically underexamined, creating a significant gap in our understanding of how these technologies may harm patients and undermine public trust in healthcare systems. This study addresses the ethical implications of AI technologies by systematically analyzing public repositories and databases that document health-related incidents. We identified 15 potential public databases and repositories, with 8 containing 488 health-related AI incidents. After removing duplicates, 295 unique incidents were analyzed, primarily from the US and UK, spanning 2012 to 2025 and affecting diverse health domains, including cardiology, COVID-19, surgery or cancer among others. Incidents were categorized into risk types, including bias, privacy violations, and misinformation, and compared with five established AI risk frameworks. The identified incidents were reflected from the lenses of the Digital Ethics Canvas. Our analysis reveals ethical gaps in current AI systems in healthcare. The apparently low number of reported incidents may suggest underreporting, potentially raising fundamental questions about moral responsibility, professional duty, and the right of patients and society to understand AI-related risks. Our findings highlight the urgent need for a more detailed assessment and surveillance of AI incidents and their impact to enhance surveillance, improve AI safety, and inform evidence-based policymaking in healthcare.