Who is the Patient Trusting? A Scoping Review of Patient-Physician Trust in the Generative Artificial Intelligence (GenAI) Era

Patients extend trust to physicians whose expertise they cannot fully inspect. Generative artificial intelligence (GenAI) is now part of that exchange. Clinicians use it, organizations deploy it, and patients bring it to appointments themselves. This scoping review asks what is known about GenAI and trust in the patient-physician relationship. Following PRISMA-ScR guidance, we searched PubMed, Scopus, CINAHL, and APA PsycINFO on March 9, 2026, and screened records in Covidence. Twelve studies met inclusion criteria. For each article, we documented how trust was conceptualized and measured, then classified each study by construct fidelity and closeness to an actual care relationship. None explicitly defined trust or used a validated trust instrument, and only one assessed trust during actual use. Most evidence came from vignettes and simulated encounters. We identified four themes. Trust varied by task complexity and risk, and trust in GenAI moved with trust in the clinician using it. Perceived empathy and communication style shaped trust judgments, as did institutional conditions. No study directly measured trust in the healthcare organization. To address this gap, we propose a triadic structure treating the clinician, the GenAI tool, and the deploying organization as three distinct objects of trust. The triad is a conceptual proposal generated by the review rather than a conclusion the included evidence demonstrates. If the organization is the most significant object of trust in the patient experience, as we suggest, making GenAI-enabled care trustworthy will require organizations to own disclosure of GenAI use and to clarify accountability to patients.