Patients Must Lead Medical AI Governance

Nature, in an editorial published online on September 8, 2026, advocates for a fundamental shift in the governance of artificial intelligence (AI) within the medical field, asserting that patients must be placed at the absolute center of these frameworks. The publication argues that current approaches to AI development and deployment in healthcare often overlook the unique needs, values, and experiences of the individuals who will be most directly impacted by these technologies. This patient-centric approach is presented not merely as an ethical consideration but as a prerequisite for building trust and ensuring the equitable and effective integration of AI into clinical practice.

The editorial highlights several key areas where patient involvement is crucial. Firstly, in the design and development phases, patients should have a voice in defining the problems AI systems are intended to solve, ensuring that the technology addresses real-world clinical needs and patient priorities. This includes input on the types of data used for training AI models, with a focus on ensuring data diversity and representativeness to avoid exacerbating existing health disparities. Secondly, patients need to be involved in the validation and testing of medical AI, providing feedback on usability, accuracy, and potential biases from their perspective. This goes beyond technical performance metrics to encompass the human experience of interacting with AI-driven diagnostic or treatment tools.

Furthermore, the governance of medical AI must address issues of transparency and explainability. Patients have a right to understand how AI systems arrive at recommendations or decisions that affect their health. The editorial calls for clear communication strategies that demystify AI for patients, enabling them to make informed decisions about their care. This includes understanding the limitations of AI and the role of human clinicians in overseeing its application. The publication also stresses the importance of patient consent and data privacy, ensuring that individuals have control over how their health data is used to train and operate AI systems.

The editorial in Nature posits that without robust patient engagement, medical AI risks becoming a tool that further marginalizes vulnerable populations or fails to deliver on its promise of improving healthcare outcomes for all. It suggests that establishing patient advisory boards, incorporating patient representatives into regulatory bodies, and developing patient-friendly educational resources are essential steps. By prioritizing the patient perspective, the medical community can foster greater accountability, enhance the safety and efficacy of AI technologies, and ultimately build a future where AI serves to empower patients and improve their health and well-being in a just and equitable manner. The doi for this publication is 10.1038/d41586-026-02796-8.