Informed consent for medical AI

Abstract Artificial intelligence (AI) in medicine raises questions about whether clinical use of AI will negatively impact patients’ self-determination and physicians’ therapeutic autonomy within the clinical decision-making process. The process of informed consent is regarded as an important touchstone of the patient–physician relationship, wherein patient and physician autonomy converge to drive treatment decisions. Informed consent has closely related ethical and legal dimensions that impact the type, scope, and means of information provided during informed consent conversations. While medical ethics facilitates informed consent in the clinical sphere, the legal system can impose liability when those rules are violated. To investigate the potential impact of AI on informed consent, we conducted a comparative analysis of 132 legal cases from the United States, United Kingdom, and Germany, alongside clinical case studies illustrating AI applications in diagnosis, treatment planning, and procedural execution. Our findings reveal that current legal principles governing informed consent are broad enough to encompass AI-driven medical technologies. The cases indicate that disclosures surrounding AI will depend on whether AI applications represent material risks, reasonable alternatives, or experimental methods, with informed consent requirements varying based on clinical integration, risk–benefit profiles, and standard of care acceptance. Because legal assessments of informed consent have already considered the challenges posed by medical innovations, inherent uncertainties of human diseases, knowledge disparities, medical information complexity, and dynamic variables that impact individual treatment options, AI integration, while potentially posing some new challenges, should not inherently undermine the protection of autonomy offered by the process of informed consent.