Data for dollars? The ethics of trading African health data for American investment: lessons from the US–Kenya deal

Recent news reports describe a multibillion dollar health cooperation agreement between the Kenyan government and the USA, under which access to ‘de-identified and aggregated’ data from Kenyan health systems would be provided to US researchers within a legally governed framework, alongside substantial US funding commitments for healthcare infrastructure and services that are needed in Kenya.1 The arrangement has been cast, by representatives of both governments, as reasonable, efficient and legally regulated with all due protections and oversight2 (see box 1). It is also framed as mutually beneficial: Kenya gains urgently needed health investment, while the USA gets privileged access to national-level data that could enhance its biomedical research.3 This research, in turn, could ultimately yield medical discoveries and health technologies that benefit Kenyans themselves, including improved disease surveillance, diagnostics and system-level planning.4 Box 1 ### A mutually beneficial deal? Announcing the agreement in Washington in December of 2025, Kenya’s Health Cabinet Secretary, Aden Duale, argued that ‘health data is a national strategic asset’ and promised that any data transfers would remain ‘fully governed’1 by Kenya’s Digital Health Act of 2023 and its Data Protection Act of 2019, which together set out legal requirements for the processing, governance and protection of personal and health data, while also establishing individual rights and independent oversight mechanisms.2 ,29 30US officials, for their part, framed the deal as part of a shift towards ‘sustainable … government-to-government cooperation,’1 with Secretary of State, Marco Rubio, describing it as a move away from fragmented donor models and towards long-term system strengthening. Another possible benefit is improving the performance and equity of data-driven health tools, since people from Africa, including Kenya, remain under-represented in many of the datasets used to develop health data-science models.5 The exchange thus appears to be not only scientifically pragmatic or economically …