The European Health Data Space Needs AI Model Sharing Provisions

In the global race for data sovereignty and AI leadership, the European Union has adopted the Regulation on the European Health Data Space (EHDS), one of the most comprehensive health data governance initiatives worldwide. Presented as a cornerstone of Europe’s AI-driven20 innovation strategy, the EHDS establishes a framework for cross-border access and secondary use of health data. Yet, while aiming to unlock data for research and innovation, it remains silent on a crucial dimension: the governance of AI models trained on EHDS data. If left unaddressed during implementation, the EHDS will continue to overlook that AI models function as essential forms of metadata for the datasets they generate, annotate, or analyze, while risking the transformation25 of EU health data into raw material for proprietary AI models inaccessible to data-generating institutions. Addressing this blind spot requires obligations for AI model transparency and sharing, ensuring that innovations derived from EHDS data benefit European research, healthcare systems, and citizens.