This Viewpoint argues that prevailing ethics-based and compliance-oriented approaches to artificial intelligence (AI) in health are insufficient for the dynamic, context-dependent realities of contemporary AI systems. It proposes a shift toward collaborative stewardship, a model that emphasizes shared responsibility, continuous learning and meaningful stakeholder participation across the full lifecycle of AI in health.
The analysis draws on a structured synthesis of peer-reviewed studies, major international policy documents and interdisciplinary scholarship published between 2021 and 2025. Using this evidence base, the paper introduces the C-STEER framework, which outlines practical components of collaborative stewardship and maps them to key stages of the AI lifecycle.
The synthesis reveals that static ethical principles and top-down regulatory models frequently fail to account for real-world variability, equity concerns and the evolving behavior of systems. Governance approaches that combine legal, technical, organizational and participatory mechanisms, supported by continuous monitoring and local adaptation, are better positioned to build trust, enhance accountability and promote equitable outcomes.
By defining collaborative stewardship and presenting the C-STEER framework, this Viewpoint moves beyond compliance-driven governance and offers a practical, context-responsive model for responsible AI integration in health systems.
