The legal dilemma of medical artificial intelligence in China: challenges to physicians' duty to inform and a typology-based response

Abstract

The rapid advancement of artificial intelligence is profoundly reshaping clinical practice and transforming the doctor–patient relationship from a traditional physician–patient dyad to a composite structure of “physician + AI–patient”. While this structural shift can improve diagnostic and therapeutic efficiency, it also poses unprecedented challenges to the normative design and legal application of the physician's duty to inform. In traditional settings, this duty has evolved from standardization to strictness and then to substantive requirements. With the introduction of medical AI, a fourth transformation toward a typology-based approach is urgently needed. Once AI participates in medical decision-making, this composite actor disrupts the trust relationship between physicians and patients, expands the scope of the physician's duty to inform, and blurs the allocation of legal responsibility. Moreover, existing AI classification schemes are misaligned with physicians' duty to inform. This article employs doctrinal analysis, case analysis, and comparative legal research. At present, international norms tend to construct legal and ethical frameworks around the core concept of “risk prevention”. In the field of medical AI, this article proposes that—based on the current state of development—a “cooperation” concept be added on the foundation of risk prevention. On the premise that medical AI should not be recognized as a legal subject, we construct a four-level classification system centered on “decisional autonomy” and the degree of impact on the physician–patient relationship, together with a corresponding tiered accountability regime. Through a tripartite linkage mechanism—decision-making authority, duty to inform, and liability allocation—we aim to achieve a dynamic balance between technological complexity and patients' informed consent, and to promote the orderly development of medical AI within the framework of the rule of law and ethics.