Artificial intelligence in primary care: frameworks, challenges, and guardrails

Artificial intelligence (AI) is already reshaping various aspects of primary care, from documentation and triage to population health planning; however, AI implementation remains fragmented, uneven, and often poorly aligned with the realities of front-line services. In this Viewpoint, we propose a functional framework for categorising AI applications in primary care, using WHO digital health interventions taxonomy as a foundation. We argue that adopting a system-level approach enables clearer identification of implementation gaps, regulatory needs, and maturity areas. Drawing on this system-level structure, we examine the technical, ethical, and operational challenges, and propose a set of high-level principles to guide the safe, equitable, and sustainable integration of AI. We conclude by highlighting the need for strong governance and participatory approaches to ensure that AI strengthens rather than fragments the core values of primary care.