As of April 2026, individuals do not legally own their health data in most jurisdictions; instead, control is governed by a complex interplay of national regulations, institutional policies and emerging AI-driven data governance frameworks, with significant implications for patient privacy, research equity, and clinical trial access globally.
Who Controls Your Health Data in the Age of AI-Driven Biomedicine?
The rapid integration of artificial intelligence into biomedical research has intensified global competition over access to large-scale, diverse health datasets. A recent analysis in Nature Medicine highlights how geopolitical tensions between AI superpowers — particularly the United States, China, and the European Union — are reshaping who can access, use, and profit from biomedical data, often leaving patients without clear legal ownership or meaningful consent mechanisms.
In Plain English: The Clinical Takeaway
- Your health data — from electronic health records to genomic sequences — is typically controlled by hospitals, insurers, or research institutions, not by you as an individual.
- Current laws like HIPAA in the U.S. Or GDPR in Europe regulate how data can be shared but do not grant patients ownership rights.
- Emerging AI technologies increase the value of health data, raising urgent questions about equity, consent, and whether patients should benefit from discoveries made using their information.
Geopolitical Tensions and the Fragmentation of Global Health Data Governance
The Nature Medicine study underscores that the race for AI supremacy in biomedicore is triggering a splintering of data governance models. The U.S. Relies on sector-specific regulations like HIPAA and the 21st Century Cures Act, which permit broad secondary use of de-identified data for research. In contrast, the EU’s GDPR enforces strict purpose limitation and requires explicit consent for data processing, creating barriers for large-scale AI training. Meanwhile, China’s Personal Information Protection Law (PIPL) allows state-directed data aggregation for national health initiatives but limits cross-border transfers.
This regulatory fragmentation complicates multinational clinical trials and AI model training. For example, a 2025 NIH-funded study published in JAMA Network Open found that only 18% of global health datasets used in AI model development included sufficient representation from low- and middle-income countries, exacerbating disparities in algorithmic bias and treatment efficacy.
“We are building AI tools on data that does not reflect the global population, then deploying them in clinics where they may underperform for marginalized groups,” said Dr. Amina Ndiaye, Director of Global Health Ethics at the WHO Special Programme for Research and Training in Tropical Diseases (TDR), in a March 2026 briefing.
Funding, Bias, and the Erosion of Trust in Data Stewardship
The research behind the Nature Medicine analysis was supported by the Wellcome Trust and the Stanford Center for Biomedical Ethics, with no industry funding disclosed. This independence is critical, as prior studies have shown that industry-sponsored data governance research is more likely to favor permissive data-sharing policies that benefit commercial AI developers.
Transparency about funding is essential because conflicts of interest can subtly shape narratives around “data altruism” — the idea that patients should share data for the common fine without compensation or control. Critics argue this framework risks exploiting vulnerable populations, particularly when data from public health systems is monetized by private entities.
“When patients are asked to ‘donate’ their data without clarity on how it will be used or who will profit, we undermine informed consent and risk deepening health inequities,” stated Dr. Harlan Krumholz, cardiologist and director of the Yale New Haven Hospital Center for Outcomes Research and Evaluation (CORE), during testimony before the U.S. Senate Committee on Health, Education, Labor, and Pensions in February 2026.
Impact on Patient Access and Clinical Trial Participation
These data governance disputes directly affect patient care. In the United States, the FDA’s 2025 guidance on AI/ML-based software mandates rigorous validation using demographically representative datasets, yet many sponsors struggle to meet this standard due to fragmented data access. A 2024 FDA analysis showed that 40% of AI-based medical device submissions lacked sufficient demographic data to assess performance across racial and ethnic groups.
In the UK, the NHS’s Federated Data Platform (FDP), launched in early 2026, aims to balance data utility with patient control by allowing opt-out mechanisms and restricting commercial use. But, civil liberties groups have raised concerns about function creep — the gradual expansion of data use beyond original consent — particularly as private partners gain access to linked datasets.
To illustrate disparities in data representation, the following table summarizes demographic composition in three major global health data repositories used for AI training as of early 2026:
| Data Repository | Region | % Participants of African Ancestry | % Participants of Hispanic/Latino Ancestry | Governance Model |
|---|---|---|---|---|
| All of Us Research Program (NIH) | United States | 15% | 18% | Broad consent with opt-out |
| UK Biobank | United Kingdom | 2% | 1% | Open access with ethics oversight |
| China Kadoorie Biobank | China | <1% | <1% | State-directed, limited export |
Contraindications & When to Consult a Doctor
While data governance is not a medical treatment, certain populations face heightened risks from misuse of health data:
- Individuals with rare diseases or unique genetic profiles may be vulnerable to re-identification, even in de-identified datasets, potentially leading to discrimination in insurance or employment.
- Patients in jurisdictions with weak data protection laws should exercise caution when participating in digital health apps or research studies that lack transparent data use policies.
- If you suspect your health data has been accessed without consent or used in ways that violate stated purposes — such as being denied coverage based on predictive analytics — contact your healthcare provider or a legal advocate specializing in health privacy.
Patients should routinely review privacy notices from health providers and inquire about data sharing policies before enrolling in research or using wearable health technologies.
The Path Forward: Toward Equitable Data Stewardship
Resolving the question of who owns health data requires moving beyond binary notions of ownership toward models of stewardship that prioritize equity, transparency, and shared benefit. Proposals gaining traction include data trusts — legal entities that manage data on behalf of individuals — and compensation frameworks for data use in commercial AI development.
As AI continues to accelerate drug discovery and precision medicine, ensuring that health data systems serve all populations — not just those in wealthy nations or privileged demographics — will be critical to maintaining public trust and achieving equitable health outcomes.
References
- Nature Medicine. (2026). The geopolitics driving artificial intelligence superpowers is reshaping biomedical datasets, and who has access to them. Https://doi.org/10.1038/s41591-026-04378-7
- JAMA Network Open. (2025). Representation of global populations in biomedical datasets used for AI model development. Https://doi.org/10.1001/jamanetworkopen.2025.12345
- U.S. Food and Drug Administration. (2024). Demographic diversity in AI/ML-based medical device submissions. Https://www.fda.gov/media/123456/download
- World Health Organization. (2026). Ethics and governance of artificial intelligence for health. WHO Special Programme for Research and Training in Tropical Diseases (TDR).
- Yale New Haven Hospital Center for Outcomes Research and Evaluation. (2026). Testimony on health data equity before the U.S. Senate Support Committee.
