Toward an Ethical Framework for Patient-Facing Artificial Intelligence in Clinical Settings

Patient-facing artificial intelligence in clinical settings raises distinct ethical challenges that require explicit attention as patients transition from passive recipients to active users of AI-augmented healthcare. We conducted 36 qualitative interviews with patients, healthcare professionals, and AI developers using a postpartum depression prediction tool as a use case to identify ethical considerations for patient-facing AI implementation. Six interconnected themes emerged, which synthesized into a preliminary ethical framework positioning transparency as the foundational enabler, the four core principles (autonomy, beneficence, justice, privacy) in productive tension, and trust as both an emergent outcome and prerequisite for engagement. Our findings suggest that ethical implementation requires making explicit the trade-offs between principles and involving patients in deliberating context-specific tensions rather than treating ethics as a compliance checklist.