From extraction to emancipation: Rethinking health data commons governance in the Global South
Online-first printer-friendly version. See DOI (following publication) for version of record.
(2026) 79 Current Legal Problems
1. Introduction
Health data aggregation is transforming the political economy of healthcare and public health governance across the Global South. The Africa Centres for Disease Control (CDC) has positioned AI as central to Africa’s disease surveillance and healthcare delivery, with the Director of Digital Health calling for a horizontal health record that would connect every citizen and track their health over a lifetime, developing a continent-wide integrated health system built on continuous data flows.[1] Across the Atlantic, the CARICOM Secretary-General has called for the region to construct a biobank to protect Caribbean biological assets, and for AI, genomics and digital health platforms to be deployed at scale across member states.[2] Together with the Inter-American Development Bank, the Pan American Health Organization has launched the Pan-American Highway for Digital Health, which envisions an integrated health data exchange that allows individuals to carry their health data and access prescriptions, medical certificates, patient summaries and care across the Latin American and Caribbean region.[3]
As centralised health data systems expand in reach and complexity, questions of access, extraction and value become increasingly important. Health data gains more analytical and operational power through aggregation across people, time and sites. Individual clinical records become epidemiologically more meaningful when pooled, compared and reused across contexts that support population health through, for instance, surveillance and evidence generation that eludes the remit of any single entity. However, this same health data can be used to re-identify individuals when linked across sources, reveal information about relatives and household, expose sensitive information about sexual and reproductive health, and generate group-level harms through biased sampling or stigmatising inferences.[4]
Increasingly we also see that data aggregation processes are mediated by corporations, through platform providers, analytics firms, cloud infrastructures and public-private partnerships. As a result, these actors are uniquely positioned to extract the most value from health data not only because of infrastructural control but also because the compilation and interpretation of datasets generate new and multiple intellectual property rights (IPRs), shift ownership from individuals to corporations,[5] without accounting for the labour that goes into producing health data.[6] Through this process of datafication, infrastructures through which data is collected and process often place corporations in a privileged position over its future use. Public health datafication[7] thus follows wider patterns that Zuboff identifies as ‘surveillance capitalism’, where human experience is rendered as behavioural data for extraction, prediction and commercial value.[8] It also resonates with accounts of digital (data) colonialism, which show how data extraction extends the colonial project by converting social life into a resource for accumulation.[9] Without effective oversight, transnational health data regulation remains poor and inequitable, with severe consequences for the Global South.
To respond to these equity concerns policymakers and scholars have increasingly turned to data commons as an umbrella concept for organising visions of a more just governance arrangement than the status quo. the For instance, WHO’s health data governance framework and the Health Data Collaborative both invoke commons rhetoric as a basis for cross-country coordination and equitable distribution of the returns from health data sharing. Scholarly proposals have largely taken three main legal forms, namely trusts, stewardships and cooperatives.[10] Each of these forms would potentially offer different mechanisms for managing access to health data and redistributing value across communities. These proposals often promise a reorientation of the health data value flow relation away from extractivism and toward shared access, democratic stewardship and collective benefit.[11]
However, data commons rhetoric inherit at least three overlapping genealogies: (1) a legal‑historical account tying commons to European land‑use, customary rights and common‑property regimes, where ‘common’ often meant access governed by membership to a locality and enforceable rules;[12] (2) an institutional framing associated with the ‘tragedy of the commons’ and its subsequent critique, especially in Ostroom’s work, which distinguishes open access from common property and treats commons as governed systems rather than unmanaged resources. This genealogy informs concepts like ‘knowledge commons’, where questions of contribution, extraction and fair allocation are central;[13] and (3) a political-economic critique that emphasises enclosure and dispossession, state‑supported privatisation (accumulation by dispossession), and the vulnerability of commons projects to market co‑optation.[14] While these genealogies overlap, they foreground different assumptions about power and the risks posed to conceptualising a health data commons.
Bringing together scholarship on the history of global health, critical legal theory and emerging theories of digital colonialism, this article argues that, despite their appeal, these proposals fail to address how capitalist modernity is built into the language and institutions of the data commons, so they do not go far enough to change governance or challenge the linked systems of datafication and extractivism. The article begins with the problem of health data value, examining how aggregation generates value while simultaneously producing governance risks. It then examines the data commons as a legal and political concept, tracing its typological forms and the intellectual traditions from which it is derived.
Following this, the article provides a critical account of the colonial genealogies of trusts, stewardship and cooperatives as legal technologies, making a novel contribution for a theory of dispossession against which health data governance proposals that use this language ought to be evaluated. These dominant legal forms through which commons governance is proposed each carry specific colonial histories that have not been fully acknowledged in the literature. Trusts have been used by white settlers to dispossess Indigenous African communities of land through the creation of national parks. Stewardship arrangements have been deployed by cultural institutions in the Global North for decades to justify the retention of stolen African cultural non-human entities and human remains under claims of preservation and universal heritage, converting struggles over ownership and reparation into technical questions of custodianship and expertise. In parts of British Africa, cooperative policy was tied closely to agricultural modernisation, export crop discipline and indirect rule. In this way, the cooperative was part of a development apparatus that organised rural life, channelled production and stabilised the colonial political economy.[15] These associations reflect a structural tendency in which governance language promises care and continuity while depoliticising the underlying conflict over who controls value, on whose terms and under whose authority. Data commons proposals that deploy these forms without seriously considering relationality in the Global South risk reproducing the same dynamics within affected communities and subordinating Indigenous or local knowledge systems to dominant epistemologies, rather than fundamentally transforming governance relations.
To counter these problems, the article turns to three case studies of data commons initiatives in the Global South, including PharmAccess in Kenya and Tanzania, Centre for Data and Knowledge Integration for Health (CIDACS) in Brazil, and H3Africa’s Biorepository Program, reading them against the theoretical framework of dispossession to identify the conditions under which progressive possibilities in law and institutional design may be realised. It concludes with a normative contribution, setting out the legal and institutional safeguards around benefit rights, community audit powers, explainability requirements and principles of refusal that would need to be built into a genuinely decolonial health data commons to enable radically different relationships of ownership, consent, stewardship and permissible repurposing of collective health data.
2. Datafication as dispossession
Health Data are uniquely sensitive and valuable. These two qualities are inseparable. Individual medical records are deeply personal, but their strategic significance grows considerably when they are combined into broader datasets that carry strong predictive value for care delivery and future research. When clinical, genomic, administrative and epidemiological data are pooled, they enable population health surveillance, predictive modelling and AI-driven innovation at a scale.[16]
Aggregation creates distinct harms. Health data can create relational harm to those linked with individuals.[17] As data are compiled, interpreted and integrated, they undergo a legal transformation: the method and product of assemblage frequently attract new IPRs, converting a formerly shared resource into a privately owned commodity and progressively distancing it from the original data subjects and the health systems through which the data were generated. At the same time, both data subjects and health systems lose any meaningful control over the data’s future use and value.[18] AI intensifies this dynamic, as AI systems require ever-larger training datasets, creating strong commercial incentives to accelerate aggregation and to enclose the resulting assemblages behind proprietary architectures. Four recent media stories illustrate what these structural conditions mean in practice.
A. Bilateral data extraction and the America First global health strategy
In January 2025, the second Trump administration launched a review of its international aid policy and, by February 2025, had formally disbanded USAID and withdrawn from the WHO as part of a foreign policy shift aimed at placing US interests at the centre of its global engagement.[19] Departing from the US’ historic commitment to ‘global health’ surveillance through agreements such as the International Health Regulations (IHR) 2005, the administration turned to bilateral agreements with African and Latin American States to secure access to biomedical specimens and health data spanning both pandemic response and general health system records. These agreements are explicitly framed as part of an ‘America First’ global health strategy, under which the US gains near-real-time access to African and Caribbean health data, including outbreak surveillance, to advance its own health security and to open pathways for American technology companies to those data environments at the expense of Global South partners.
B. IPRs, aggregation and the black boxing of public data
The legal transformation that aggregation produces can be stark. For instance, in Kenya, a government challenge to the integrity of its own electoral data was blocked in part by IP arguments raised by the private technology company, Smartmatic, that had processed and compiled the voter tally records.[20] Once the company had aggregated, interpreted and structured the raw data into compiled form, that form attracted IP protections that rendered the underlying public record legally opaque and commercially enclosed. An entire country’s voting record became less visible to the state that commissioned it because the amalgamation process had changed its legal character and shifted the nature of the rights that the state held over it. This case also demonstrates that as data moves from collection through aggregation to compilation, it passes through a series of legal thresholds at which ownership claims multiply and deepen, progressively narrowing the field of actors who retain enforceable rights over the resource. For health data, which is still collected overwhelmingly through public health systems funded by public money, this trajectory represents a transfer of publicly generated value into private hands that is rarely acknowledged in governance debates.
C. AI, cultural data and the limits of community control
The risks posed by data aggregation and AI for communities are particularly evident in the experience of Te Hiku Media, a Māori-led organisation that pooled audio archives and developed a Kaitiakitanga-based transcription system to support language revitalisation. Māori speakers across the country recorded more than 300 hours of annotated audio in te reo Māori in less than 10 days.[21] This provided enough data to develop language technologies for te reo Māori, including automatic speech recognition and speech-to-text tools.[22] As the project scaled and AI capabilities advanced, external actors sought to repurpose the underlying models and datasets in ways that would have absorbed the Māori-led archive into broader AI language datasets, undermining the cultural integrity of the project and effectively bypassing the community governance structures built around it.
Questions of community control are amplified by what is happening in health data governance. Recently, the Gates Foundation and OpenAI launched Horizon1000, a USD 50 million pilot of an AI-driven health partnership aimed at supporting African healthcare systems, with Rwanda selected as the pilot country and a target of supporting up to 1,000 primary healthcare clinics and surrounding communities across several African countries by 2028.[23] The programme aims to work with governments and health authorities to identify ways AI could support healthcare delivery, including clinical decision support, disease surveillance and administrative efficiency, with Bill Gates describing AI as potentially important for addressing healthcare gaps. The announcement generated sceptical reactions online, with critics warning that the initiative could enclose the value of African health data from the continent.
D. Digital colonialism
The three vignettes foretell the structural problem that will shape future health data governance, especially across the Global South. They point to a set of problems that existing scholarship has increasingly theorised through the language of data colonialism, digital colonialism and algorithmic coloniality. Nothias’ intellectual history of the field shows that the starting point is usually the claim that contemporary data extraction should not be treated as merely a technical problem of privacy or cybersecurity regulation lagging; rather, it is part of a longer history in which social life, bodily vulnerability, labour and knowledge, as well as institutional dependency, are converted into sources of value for actors positioned elsewhere.[24] Couldry and Mejias’ account of data colonialism argues that the continuous capture of human life through data is not simply analogous to historical colonialism, but extends its logic by turning social relations into raw material for capitalist accumulation. For them, the distinctive feature of the present is not territorial occupation but the creation of infrastructures through which ordinary life becomes available for extraction.[25]
A related body of literature on digital colonialism shifts attention from data capture alone to the wider political economy of digital infrastructure. Kwet’s work on South Africa is especially important because it shows how US technology corporations exercise power by both collecting data and dominating the underlying architecture of digital life, including software, hardware, cloud systems and network connectivity, producing a form of techno-imperialism that concentrates economic and epistemic power in the Global North while extracting value from Southern markets and populations.[26] Extraction is not a singular moment of data collection but is built into the infrastructures through which data is collected and interpreted, and made useful. Even where formal sovereignty or public ownership remains intact, dependence on foreign technological systems can shift practical control over data’s movement, interpretation and value.
Scholarship on algorithmic coloniality extends this critique to the design and deployment of AI systems. Birhane and others show how that algorithmic systems often reproduce hierarchies of race, gender, geography and knowledge even though they are often narrated as universal. Beyond being biased because of bad data, the problem is that the assumptions embedded in training data, model design, benchmarking practices and deployment contexts reflect the social and epistemic location of those who build and govern these systems.[27] When such systems are exported into Global South contexts, they may carry with them the priorities, classifications and interests of the actors that designed them, while marginalising local knowledge, and situated and alternative understandings, of health and risk.
What these important interventions show is that contemporary digital systems reproduce colonial relations through data extraction, infrastructural dependence and epistemic hierarchy, which explains why the language of innovation and development often structures the introduction of these technologies. Technology corporations present digital systems as solutions to problems like inefficiency and weak institutional capacity but then the same systems deepen dependency by making public institutions reliant on proprietary infrastructures and capacities concentrated in the Global North. In this way, the promise of technological progress is animated by the logic of benevolent assistance while reorganising social and institutional life around extractive value production.
In addition, this technology comes to public health systems in the Global South that are underwritten by a history of colonial medicine that functioned simultaneously as a mechanism of control over colonial subjects and as a mode of extraction in which Indigenous plants, health knowledge and biomedical materials were appropriated for the benefit of imperial powers.[28] Law was central to this process. Colonial legal regimes developed regulatory powers over how medicine could be practised, created new property forms that could enclose Indigenous knowledge, and later deployed instruments such as quarantine to restrict the movement of colonised populations in ways that served imperial administrative and economic interests rather than the health of the people subjected to them.[29] The production of health knowledge was therefore shaped by legal structures that determined whose bodies were objects of investigation and whose knowledge systems were recognised as legitimate.
E. Towards datafication as dispossession
Returning to the three vignettes, it is apparent that health data is generated through social life and labour, including through care, illness, clinical labour, disease surveillance, public health research, administration and community life. Yet once these relations are translated into data, they can be aggregated, interpreted and circulated through infrastructures controlled by actors far removed from the people and systems that produced them. The literature on data colonialism helps us see this as extractive. The literature on digital colonialism helps us see the infrastructural conditions that make extraction possible. The literature on algorithmic coloniality helps us see how technical systems reproduce epistemic hierarchies.
Data can therefore be understood as the commodified or assetised form of human experience. Yet value extraction from health data is mediated through the same process through which human experience becomes enclosed from communities. Mechanisms through which this erosion occurs vary, from bilateral agreements that formalise access based on IPRs, to processes that assetise data, to AI deployment. Datafication as dispossession enables us to see the process through which embodied and social experiences of health are transformed into forms that can be accumulated and enclosed by others. Relations of care, illness, public health labour and human biological make up are recorded and take on a new form, but when those relations are aggregated and interpreted both the process of aggregation and the infrastructure through which these processes happen create new legal forms, which encloses the value of human experience from the very communities whose experience it is.
The first mechanism of dispossession is legal. Bilateral agreements – just as data-sharing agreements, which are common in health, secure access to health data – formalise unequal relations through legal instruments that look reciprocal but often operate within asymmetric conditions. Data-rich but resource-constrained health systems in the Global South become suppliers to data-processing capacities concentrated in the Global North. These arrangements may be framed as scientific exchange or mutual benefit, but their distributive structure is frequently uneven. The communities whose health records anchor the arrangement often have limited control over secondary uses, limited legal recourse when data is repurposed, and limited claims over the value generated from their contribution. The dispossession lies not only in the transfer of data but in the legal ordering of that transfer as legitimate cooperation while the benefits of processing, modelling and commercialisation accrue elsewhere.
The second mechanism is proprietary mediation. The Kenyan election case illustrates a broader dynamic central to digital governance. Public data can become less accessible when the infrastructures required to store, transmit, authenticate, interpret or reproduce it are privately controlled. In that case, the difficulty arose because access to those records was mediated through proprietary software, security systems, encryption keys, transmission certificates and technical processes protected by IPRs, not because the underlying voting records were themselves private. Data does not need to be privately owned at the point of collection for control over it to shift. Control can be relocated through the software, databases, AI models, standards, platforms and security architectures that determine how data is made meaningful and usable.
The third mechanism is assetisation. As health data moves from collection through various stages of aggregation, compilation, interpretation and modelling, it passes through legal and technical thresholds at which new rights and claims can attach. An individual’s clinical record or genomic sample or an epidemiological dataset may begin as part of a public health function. Once aggregated and cleaned and embedded in an AI system, it may become part of a protected database, proprietary model, trade secret, copyrighted compilation or contractually restricted asset. At each stage, the field of actors with enforceable control may narrow. Patients, communities, clinicians and public health systems may remain rhetorically central as sources or beneficiaries, but the rights that govern the resulting data products increasingly attach to those who process, structure and commercialise them.
The fourth mechanism is epistemic displacement. Datafication reorganises how health is known. Health experience becomes legible through categories, standards and analytic priorities designed by institutions that may be far removed from the people whose lives are being rendered into data. This is especially significant in Global South health systems, where external actors often control technical infrastructure and analytic capacity. The risk is that local meanings of health are displaced by the classificatory systems of donors, platforms, technology companies or global health agencies. Dispossession is therefore epistemic as well as economic. It concerns who has authority to determine what the data means.
The fifth mechanism is temporal capture. Health data collected today may not produce its most valuable uses until years or decades later. A dataset may appear marginal at the time of collection but later become valuable because of new AI techniques, genomic tools, outbreak risks, pharmaceutical markets or surveillance needs. This temporal lag creates a structural problem for consent and benefit-sharing. Communities cannot easily assess the value of their contribution, and institutions cannot fully anticipate the technologies or markets that will later make the data valuable. Governance frameworks built around present uses designed to protect their interests may be entirely inadequate to the institutions and technologies that eventually exploit that value. The COVID-19 vaccine, for instance, was developed through research that relied on decades of prior health data. Other examples, such as the genomic sequencing data collected during the Ebola crisis show that there was little benefit for most people in the Global South who contributed it, and the vaccines and treatments developed from that data flowed primarily to institutions in the Global North.[30]
Two features specific to health data compound the problem. First, health data is relational and aggregative. It concerns individuals but may reveal information about relatives, households and communities, and therefore challenges the notion of informed consent, which has always been foundational to the delivery of healthcare. It is often very difficult to get patients to give informed consent to data use and account for future harms that they themselves and clinicians cannot always clearly envision. Even if patients could consent to future secondary uses through aggregation, data takes on a new form, giving rise to IPRs, which weaken the link that individuals who consented to its use have to its ownership. Individual data rights are not well suited to governing relational, aggregative or intergenerational data flows through consent agreements.[31] This explains why the long-standing tradition of surveillance-based collection of data has always operated with minimal individual consent and relied instead on the promise of anonymity to reduce risks to the individual.[32] With deanonymisation through datafication, this promise of anonymity has become weaker.
Second, health data is institutionally embedded. It is produced through public health systems, clinical infrastructures, surveillance programmes, laboratories, hospitals and community health workers. National health systems routinely collect vaccine records, maternal mortality data, disease surveillance data and hospitalisation statistics without individual consent, on the basis that anonymisation removes the individual risk and that the public health benefit justifies the trade-off. That logic was coherent when the link between the data and the health system that collected it was direct and clearly bounded. It becomes far more contestable as datasets are integrated with administrative, economic and other kinds of social data. Dispossession does not affect only individuals but may also weaken the authority of public health systems themselves. A state or health institution may collect the data but still lose meaningful control over its future interpretation, reuse, commercialisation or integration into proprietary systems. The problem is therefore not solved by locating formal ownership in the state, nor by relying on individual data rights. The issue is the social organisation of value across the full life cycle of health data.
Centring coloniality within this analysis enables us to see these dynamics not as technical failures or policy gaps but as structural features of a global health data economy that was never designed to distribute its benefits equitably. Yet both WHO and Africa CDC have embraced the aggregative vision with considerable enthusiasm. The WHO’s recent framing of health data governance in the age of AI explicitly promotes digital health infrastructure designed to generate data for AI and machine learning, expressing confidence in the capacity of large integrated datasets to drive health outcomes on the continent.[33] A decolonial approach requires attending to how power, knowledge hierarchies, and inequality are inscribed in digital health technologies and health systems, and it demands that alternatives be built from epistemic resources generated in and by the Global South rather than imported from the institutional traditions that produced the problem.[34] The data commons, in this context, is important not because it offers a ready-made solution but because it opens a conceptual space in which questions of ownership, control and benefit can be asked differently. Whether that space is genuinely transformative or is quickly closed off by the legal forms and capitalist relations through which commons projects are constituted is the central question this article pursues.
3. Imagining resistance: what could the data commons offer?
The colonial genealogy of health data governance traced above does not yield a simple conclusion that law and institutional design are irredeemably compromised. It does, however, demand that any governance alternative be evaluated against that genealogy rather than in isolation from it. The data commons has emerged as the most widely discussed structural alternative to extractive health data governance, promising to reorient ownership, access and benefit-sharing around collective rather than proprietary logics. Before that promise can be seriously assessed, it is necessary to examine what the data commons means as a legal, institutional and political concept, as the term carries significantly different implications depending on which intellectual tradition is invoked and which institutional form is proposed.
A. Typology, contestations and possibilities for data commons
The term data commons is used in at least three different ways in the literature. Some scholarship uses it descriptively for technology infrastructures that host shareable datasets and tools. The US NCI Cancer Research Data Commons (CRDC), for example, describes data commons as structured repositories that house NCI-funded research data and provide search, access and reuse features, with managed standards and harmonisation practices, and a mix of open- and controlled-access data.[35]
A second usage is explicitly institutional and theoretical, drawing on natural resource commons theory and adapting it to data and knowledge.[36] Elinor Ostrom’s work has been pivotal in exploring how communities govern shared resources through rules, boundaries and collective action arrangements.[37] Knowledge commons frameworks adapt this tradition to information, data and knowledge goods, defining the commons as an institutionalised sharing arrangement among members of a community, which enables members to decide on the most effective governance model.[38] Ostrom’s later work with Hess was foundational, arguing that digital knowledge creates unprecedented access possibilities while simultaneously facing threats from expanding IPRs, over-patenting and inadequate preservation, and that commons governance frameworks offer a way of addressing these pressures through shared rules and collective stewardship.[39] A structured knowledge commons framework adapts Ostrom’s institutional analysis to the distinctive attributes of information goods, in medicine and health, demonstrating how researchers, clinicians and patients can generate and share innovations through self-governing arrangements without defaulting to traditional intellectual property logics.[40] However, a common-pool resource framework cannot be straightforwardly applied to health data because health and personal data categories are too dynamic and unstable to constitute a bounded commons resource in Ostrom’s sense. Instead data ecosystems rather than data themselves should be treated as the relevant commons unit and any workable governance model must carefully attend to which stakeholders are affected by data practices and how governing institutions can build the communication and trust necessary to produce enforceable rules for sustainable data use.[41]
A third, more radical, usage is political, invoking communing as a project of democratising control over data and resisting enclosure by states or firms.[42] Here, the commons is not primarily a governance architecture but a claim about collective self-determination and the forms of community authority that market enclosure forecloses. This usage draws more directly on traditions of political economy, resistance and social movements, and it is more attentive to questions of power, dispossession and the structural conditions under which genuinely democratic governance is possible. It is this third tradition that the article returns to when evaluating what a genuinely decolonial health data commons would need to look like.
B. Health data and the commons as a political project
Some scholars have advanced an alternative framing of data as riparian in order to account for the relationality in order to help account for how one actor’s collection or processing of data can harm or benefit others and how cross‑border flows create transnational effects.[43] Riparian language or the language of ‘Data Rivers’ is important as it can illuminate ecological conditions of circulation which rely on continuing flows of human produced material and yet the resulting benefits are unevenly distributed both nationally and transnationally.[44] Attention to this riparian framing enables us to focus on long-term obligations, ecosystem impacts and intergenerational responsibilities of data governance but commons language better names the social relations, authority structures and design choices that the paper seeks to interrogate. Therefore, given the decolonial imagination this article aims to advance, which requires centring epistemic authority,[45] it focuses on the commons framework because of its focus on the political and institutional work of collective governance, namely who forms governance bodies, who sets rules, who benefits or is excluded?
Starting from the position that the health data commons is a political project illuminates the tensions in attempting to address entrenched coordination problems by establishing shared governance, standardised data practices and transparent access rules that align diverse stakeholders towards common goals. By pooling interoperable health data under democratic oversight, a commons enables collective benefit through accelerating research, improving public health responses and ensuring equitable distribution of insights and resources. Such institutionalised collaboration can also potentially create economic and social value by reducing duplication of effort and unlocking new discoveries at scale, provided the commons embeds safeguards that build public trust and accountability.[46]
Health data relies on pooling secondary data as an infrastructural resource with three significant features for the commons.[47] Data is a non-rivalrous good, in that it can be used by multiple users without depleting it. The same dataset can be used concurrently by researchers, clinicians, policymakers and innovators without being consumed or diminished. This characteristic enables widespread, multiplicative benefits, accelerating discovery, improving care and informing public health decisions. However, non‑rivalry does not eliminate competitive behaviour in which private entities may seek to enclose data to secure exclusive use of it and prevent competitors from using it to drive up its value. Enclosing data also leads to social impacts, especially if the public feels they have not gained value from their data. Trust, consent and equitable benefit-sharing shape who can use the data and to what ends, and temporal dynamics mean that harms and gains may emerge long after initial collection. Although collected largely by health systems, it is often difficult to imagine and envisage what benefits and, in some cases, harms will arise from them. This creates problems for communing of health data.[48] For instance, the UK Biobank, which is the world’s biggest biomedical research resource that holds genetic, health and lifestyle data on 500,000 British volunteers, has faced not only serious criticism due to the lack of transparency in consent measures around data sharing and risks around reidentification but also questions around organisations that were allowed to access data to pursue a pseudo-racist agenda.[49]
Health data commons are best understood in the literature as an adaptation of knowledge commons thinking. They are normally situated in contemporary infrastructures for biomedical research, clinical informatics and digital health platforms. This framing enables us to shift our understanding from scarce, rivalrous natural resources to information environments where exclusion, coordination and institutional design matter.[50] The commons can be seen as an alternative mode of governance because it addresses how digital knowledge offers unprecedented access while also being threatened by IP laws, over-patenting and a lack of preservation.[51] The recurring claim in this literature is that enclosure pressures have now moved from land and forests to information and culture, producing a wave of second enclosures for intangibles and public-domain-like resources.[52] In this way, expanded IP rights lead to increased fencing off of parts of the intangible creative commons that have now moved into scientific and technical forms.[53] Therefore thinking about data commons in health challenges us to frame health data governance as a contest over enclosure versus shared infrastructures for research and public benefit.[54] Critically, this poses the question about whether IP rights over health data should be restricted by law so that they are time limited, justified by reference to a public good and revokable if they fail to serve the public interest.
In this way, data commons could enable us to mitigate between boundary problems and social dilemmas, including how to define the resource and community that will accrue benefits when the data are personal, relational and reused across contexts.[55] What this framing tends to understate, however, is that these economic incentives do not operate as neutral market pressures. Capital actively uses law to convert shared resources into privately owned assets, through property, contract, collateral, trust and corporate, generating and concentrating wealth while externalising risk onto the communities from which the underlying value was extracted.[56] Health data commons proposals that treat legal regimes and economic incentives as external constraints to be navigated, rather than as the mechanisms through which enclosure is actively produced, risk underestimating the structural pressure they face and overestimating the resilience of governance arrangements that have not been designed with those pressures explicitly in view.
Despite these challenges, the enduring appeal of commons language lies in what it promises at the level of governance orientation, which is a shift away from exclusive control and towards shared access, mutual obligations and collective benefit. In health contexts, this appeal has driven a range of pragmatic initiatives that use commons principles to describe data infrastructures intended to enable research, service planning and innovation. DHIS2, the open-source District Health Information Software maintained by the Health Information Systems Programme and deployed across more than 100 countries in the Global South, functions as a commons-like infrastructure in precisely this sense.[57] Collectively developed, freely available and governed through a distributed network of national implementers, it embodies the aspirations of shared access and democratic stewardship that commons advocates describe.[58] Yet DHIS2 also illustrates a structural limit that runs through all such pragmatic commons arrangements. The platform itself may operate as a commons resource, but the governance of the data collected through it remains fragmented across national legal regimes, donor priorities and proprietary analytic tools that sit atop the open infrastructure, reproducing extraction at the data layer even where the infrastructure layer is genuinely shared.
Additionally, health data commons have a further structural problem. Unlike other commons infrastructures, which tend to operate as relatively integrated systems, health data commons span multiple technical and legal layers, each with distinct ownership structures, capital relations and governance arrangements.[59] The physical infrastructure of servers, cables and network hardware; the software platforms and application layers through which data are collected and processed; the data governance protocols and access control mechanisms; and the data itself as a common good are all potentially held under different legal forms, by different actors and on different terms.[60] Each of these layers can be independently sold, traded, licensed or terminated, meaning that a commons governance arrangement at the data layer provides no security against transformations at the infrastructure layer that can render the commons inaccessible or redirect its value entirely. This vulnerability is particularly acute in contexts of philanthropic dependence, where the continuation of infrastructure is contingent on the strategic priorities of funders who are not accountable to the communities the commons is meant to serve. It also arises in state-managed systems that rely on proprietary software whose licensing terms can be altered unilaterally, and in commons projects that are hosted on commercial cloud infrastructure whose terms of service can change without notice. Any serious account of health data commons governance must therefore attend not only to the governance of data as a common good but also to infrastructure ownership and the capital relations that govern it.
4. Data commons and coloniality
Indigenous and decolonial scholarship raises questions about using the commons as a descriptive term without paying sufficient attention to the political landscape or epistemic histories. For instance, McCormack’s important work on Māori fisheries argued that commons proposals had failed to engage adequately with inequality, that the mere use of the term commons didn’t create redress for existing property struggles, and that it may just reinforce a neoliberal environmental capitalist logic.[61] The idea of creating collective value without confronting particular historical legacies has also been found problematic in work on relational commons in settler-colonial contexts. Commons arguments must not be used to mobilise collective data arrangements without resolving questions of value, extraction and control, especially in instances where data infrastructures are dependent on platform infrastructures that reinforce colonial hierarchies.
Given that data is often pooled, aggregated and transformed through innovations to gain the greatest value,[62] there have been significant debates about the best ways to do this collectively to maximise value while reducing duplication costs.[63] This pooling of data could take three distinct legal forms: trusts, stewardship and cooperatives. A trust is a model of governance that enables groups to pool their data rights through an intermediary known as a data trustee to leverage those rights. This could include the data trustee having the authority to allow access to, aggregation of and use of the data on behalf of the group. The data trustee may then be able to negotiate better terms and conditions with service providers and/or monitor data sharing agreements.[64] Data trusts can hold public data, integrate data from multiple sources to inform policymaking, promote innovation and address societal challenges, while adopting a responsible approach to the use of personal data.[65]
Data stewardship is widely used to refer to tasks and activities pertaining to the management of commonly held data. Most understandings converge on the need to ensure responsible data practices, but there are different conceptions of the nature of responsibilities that lie at the heart of data stewardship and of which stakeholders should benefit from the data.[66] Data stewardship is best understood as a governance model that assigns continuing responsibility for data across its lifecycle, rather than treating data as a one-off byproduct of service delivery or research.[67] In the health governance literature, data stewardship is rooted in fair information practice and focused on the management of identifiable data through methods of acquisition, storage, aggregation and deidentification, alongside procedures for release and use that attend to privacy, security and ethics.[68]
In public data infrastructure, the language of data stewardship shifts from a fiduciary responsibility to a focus on relationality through working with communities towards a public benefit while taking the needs and interests of data subjects or beneficiaries into account.[69] This could be through the steward making a decision or through a bottom-up stewardship arrangement in which data subjects take a more active role around the specific uses of data. This could pose challenges in areas where data is retained for long periods, as the links between data subjects and the data may weaken. It is also a problem in highly technical areas where data subjects may not have a clear understanding of how the data will be used.[70] This is a particular problem in many jurisdictions in the Global South, where data protection acts largely delegate these functions to individual data holders.
To respond to this, scholars are pushing for the professionalisation of the role of the data steward, which could also integrate other normative processes that lead to new ways of working or standards of practice. For instance, some have argued that data stewards should occupy a fiduciary trust-like role in relation to the data and to those whose interests are affected by its use.[71] Legal scholarship and policy work on data trusts and data stewardship have developed this idea further by identifying fiduciary duties, intermediary governance and purpose-driven decision-making as mechanisms through which stewardship can be made accountable.[72] Participatory data stewardship visions, particularly from the Aapti Institute in India, add an important Global South-facing dimension by framing stewardship as a process designed to empower individuals and communities while protecting rights. This firmly puts the societal value of data within an empowerment framework.[73] African data governance scholarship and policy frameworks also push for the language of digital sovereignty, institutional capacity, data self-determination, communal accountability, and investment in data institutions and infrastructures.[74]
Although similar to data stewardship, data cooperatives enable a higher involvement of their members in the day to day decision-making through democratic controls that allow them a more direct say in the running of the cooperative, which also often results in demands for an equitable share of the benefits produced from the data.[75] In health, this retention of control is challenged by the members’ lack of technical competence. The Liverpool City Region Civic Data Cooperative (LCR CDC), while not being an actual cooperative within UK law, illustrates how this is feasible. As a civic place-based organisation, it brings together or aligns multi‑institutional data flows with public values and has been visibly experimenting with new participatory forms of oversight.[76] The LCR CDC has engaged with over 150,000 residents and supported over 20 public health and innovation projects using a newly created regional NHS secure Data Environment.
Each of the three legal forms this section has examined – trusts, stewardship arrangements and cooperatives – has been proposed as a vehicle for governing health data in ways that give communities more meaningful control over the value their data produces. But treating these forms as interchangeable technical options for selection on grounds of operational fit misses something important. Legal forms are not neutral. They carry histories, and those histories matter particularly in Global South contexts where many of the same forms were deployed as instruments of colonial administration. Attending to these histories doesn’t mean that law is irredeemably tainted or that legal forms cannot be repurposed towards different ends. Rather, this argument enables us to critique not only whether any proposed form of commons governance works in principle but also what work it has done before, in whose interests and at whose expense.
Trusts have a particular history in international law and colonial administration, where they were used to manage land, states and items of significant cultural value on behalf of populations deemed incapable of governing their own affairs. The League of Nations Mandate system formalised this logic under Article 22, designating certain colonised territories as a ‘sacred trust of civilisation’ to be administered by imperial powers until their peoples were judged ready for self-governance, an arrangement that combined a language of care and stewardship with the practical reality of continued dispossession and extraction.[77] In Kenya, trust land was used to hold land nominally belonging to Indigenous Kenyans in trust for their benefit, while the colonial administration allocated the most productive land to white settlers under the Crown Lands Ordinance of 1902 and subsequent legislation, a process through which trusteeship was converted from a protective mechanism into an instrument for legitimising land alienation.[78] More recently, self-organising groups have set up trusts that place their legal rights in an independent trustee. The trustee must act solely in the group’s interest, is supervised by a court, and must prove they have not breached their duties. Therefore, the argument this article advances is not that trusts by their nature are colonial, but, in thinking about trusts as particular legal technology with a particular history in the Global South, we can ask critical questions about how communities from the Global South can constitute fiduciary relationships from the bottom up in health data trusts. This would involve thinking carefully about the role of trustees, who must come from those communities, and setting the trusts’ terms so that these serve a decolonial agenda.
Likewise, the language of stewardship often bundles together ethics, technical practice, institutional governance and accountability, which makes it attractive but also prompts us to reflect on the relational nature of the ultimate beneficiaries of these legally instituted relationships. In Africa, stewardship has a long history with cultural artefacts. Select museums in the Global North have often justified holding cultural objects, human remains, and archives through narratives of preservation and universal heritage, presenting themselves as responsible stewards rather than possessors.[79] Contemporary museum scholarship, for instance, shows how this language sometimes obscures the violence through which collections were acquired and can reframe restitution as a threat to preservation instead of making a strong case for an obligation for reparations.[80] This history links to health data governance in profound ways. Health data are also collected, amalgamated and often curated to be made legible and useful to institutions. The language of stewardship can legitimise paternalistic extractive practices by institutions such as universities, corporations, health foundations and other states, which obscures the control that they wield through an illusory language of care.
Stewardship and trusteeship often promise care and continuity. In the past, both terms have depoliticised conflict by reframing struggles over land, culture and life as technical questions requiring expertise. In health, these struggles against medicalised expertise are not new and fields such as bioethics exist to challenge these expert-led notions of providing health care. However, the critical question remains, when we advocate for data stewards within data commons, who gets to steward, stewardship of what, for whose benefit and under whose authority? Adopting coloniality as a transformational lens enables us to embark on a decolonial project to consider what resistance in data commons would look like. One way of thinking about this is to engage with the ways in which health data commons are currently working in the Global South in an attempt to use epistemic knowledge from the South to contribute to ongoing debates on the ways in which data commons could enable communities and users to have more power over the use of their health data.
5. Lessons from data commons in the Global South
There are various forms of data commons situated in the Global South, ranging from national and transnational commons projects that use Open Science to promote specific goals like universal health coverage (UHC), such as PharmAccess, to country projects such as CIDACS. Data commons also include surveillance research projects that aim to build capacity, such as the Biorepository Program, which was part of the H3Africa consortium and helped build genomic capacity across the African continent. Additionally, there are examples of speculative commons in which Global South scholars may use the commons for normative projects in public health, for instance a feminist health data commons.[81] There are sometimes overlaps, as commons aimed at delivering services or encouraging the development of innovative products may also have a strong research base to build capacity, but they may also have a speculative normative agenda that has not yet been realised. These examples illustrate how Global South actors are experimenting with commons-like architectures under resource constraints and amid strong concerns about data extraction and inequality. The examples also enable us to draw lessons about potential harms explicitly linked to their regulation in Global South contexts.
A. Co-optation through partnerships for public good
PharmAccess was conceived as a test bed for a health data commons project in the Global South to demonstrate how access to patient health data in sub-Saharan Africa could be an essential pillar in delivering UHC.[82] The PharmAccess project works with expectant mothers through an app called Mom Care to incorporate access to financing maternal care, encouraging quality standards for healthcare providers, and creating and sharing actionable data with patients and practitioners about pregnancy. The project has shown how patient‑focused health data can be shared and reused in sub‑Saharan Africa at a reasonable cost and with low technical risk. It builds on systems already widely used in Africa and uses a common standard, i.e. mothers received a health wallet through a digital payment app (M-TIBA) to pay for health care.[83] The project aims to link different kinds of data, such as shared records and registers, to different front‑end tools, such as clinic software and apps or mobile payment apps.[84] As a health data commons, PharmAccess aims to collectively reuse data for public interest under a normative agenda of data solidarity for the benefit of patients and vulnerable groups.
Despite these aspirations, the governance of the PharmAccess project in Kenya and Tanzania remains largely technical, enabling data sharing across payers and providers rather than fostering direct, collective decision-making power for patients or communities.[85] In addition, given its desire to engage with private partners, it encourages openness of data sets in ways that generate proprietary value from formerly owned commons data.[86] This is in tension with its other aspiration around common wealth licensing, which calls for licences to ensure that value is shared by the community.[87] The lack of collective decision-making also creates an underlying accountability issue. As it has scaled up, so has the involvement of insurers and financiers, who operate in opaque ways and raise the risk of more data being enclosed without accountability.
B. Naming data commons as a form of extraction
There is a proliferation of the term data commons in the context of open-source data infrastructures provided by private corporations, although these may not be health data commons in the strict sense used in recent governance debates.[88] Google’s General Data Commons, for instance, aggregates public statistics (including health indicators) for countries like Brazil, India and South Africa, with visualisation tools.[89] While this infrastructure lowers barriers to accessing comparative health statistics, it does not provide participatory governance or local control mechanisms over how data are integrated, repurposed or used to train commercial AI systems. It is better understood as an open data infrastructure that operates under Google’s own governance terms rather than as a health data commons in any meaningful institutional sense.[90]
The distinction is important because open data infrastructure and democratically governed commons are not interchangeable and treating them as such obscures a structural asymmetry that runs through Google’s broader health data engagements. Couldry and Mejias describe this as data colonialism’s double logic, in which data resources are extracted and transformed into private capital with minimal benefit flowing back to the communities that provided them.[91] More broadly, corporations training AI models for health are increasingly sourcing electronic health records from the Global South, where data has historically been difficult to collate, and where regulatory frameworks are weaker and consent mechanisms thinner, with the resulting AI infrastructure often protected by IP rights and stored on proprietary servers that Global South institutions cannot independently access or leverage.[92] In this context, the infrastructure that aggregates Global South health data under a commons label while concentrating analytical capacity and commercial value in the Global North reproduces the extractive logic the commons is supposed to resist, and does so with the added problem of terminological cover that makes the extraction harder to name.
C. Data localisation and the commons conundrum
The historical export of biospecimens from Africa, highlighted during crises like Ebola, has driven calls for data localisation and sovereignty governance mechanisms such as Kenya’s Digital Health Act.[93] However, keeping data physically in-country does not resolve the core data commons dilemmas of value and harm that data can still be copied or accessed remotely, so extraction of value through data may persist while harms such as privacy breaches and inequitable benefits remain possible. Localisation of commons, especially when managed by the state, requires long-term sustainability plans, including budgets, capacity building and strong political will, which is often challenging in neoliberal contexts where the market sells efficiency to the state. A stronger commons argument, therefore, must prioritise enforceable access controls and equitable benefit‑sharing arrangements, emphasising that protections should shift from mechanisms of transfer to a focus on the value of data itself, ensuring local community participation and valuation mechanisms, alongside regulation to identify, quantify and remediate harms over time.
D. Beyond money: Indigenous CARE Principles and questions of value
The CARE Principles – collective benefit, authority to control, responsibility and ethics – were designed to centre people and purpose over purely data‑centric reuse, and by subjecting the idea of stewardship to evaluation against social outcomes rather than technical metrics alone.[94] This framing helps stewardship escape colonial and neoliberal legacies by insisting that governance deliver real benefits and legitimate authority for communities. Evidence of value should therefore be broadened beyond immediate economic gains; for example, clinician time saved, cost reductions, or commercialised innovation. It should include social and cultural returns through measurable community health improvements, locally governed capacity building, and relational or intangible values documented in Indigenous commons literature, such as reciprocity, knowledge sovereignty and strengthened communal autonomy.[95] It is true that these values are harder to measure. However, there is excellent work from environmental law that attempts to create impact assessments for intangible elements such as intergenerational harm.[96]
E. Capacity training projects H3 Africa
The H3Africa Biorepository Program, which could be described as a commons, hosts biospecimen collections from multiple African countries. It not only centralises valuable samples for scientific use that benefit the African continent, through more stringent benefit-sharing agreements that illustrate how the use of these samples will benefit Africans more directly, but also cultivates capacity for African scientists. As part of the large, international H3Africa collaboration, this programme was transformative in developing a new generation of health data stewards who now interrogate research practices, governance and ethical use of health data, and who help ensure that research can serve local priorities as well as global science.[97] This focus on capacity training is vitally important for future data commons projects.
F. Reparatory project: the case of CIDACS
Behind data commons projects must be an underlying rationale for reducing inequalities in truly transformative ways. CIDACS, also known as the 100 million Brazilian cohort, is a good example of this. Established in 2016 as part of the Oswaldo Cruz Foundation, it functions as a data common by linking Brazil’s large administrative and health databases into curated, interoperable resources that support population‑scale research and policy evaluation. CIDACS’ infrastructure, governance practices and secure linkage methods enable scalable, non‑rivalrous use of data while foregrounding ethical oversight, reproducibility, and use for social determinants and equity‑focused research, thereby turning formerly dispersed administrative records into shared public value.[98]
In trying to link epidemiology and science to contemporary concerns about reparations, equity and social determinants of health, data commons such as CIDACS can (i) produce new knowledge through epidemiological data in order to create policies that both address historical injustices (land, labour and racial exclusions) and redistribute resources through linkages with social security benefits and (ii) prioritise community-led measures that improve health and other socioeconomic inequalities.[99] This is achieved through a stringent oversight mechanism that supports evidence-based policy making while also protecting data subjects’ rights. It also involves ensuring that research collaborations include local partnerships that understand the context, doing more to find out the priorities of underrepresented communities to ensure that research priorities are targeted to these needs. A 2020 study on CIDACS’ use of administrative data has, for instance, expanded the remit of CIDACS into broader public engagement.[100]
G. Engendering participation
To be truly effective as a knowledge commons, a data commons must ensure meaningful participation through community governance mechanisms. The idea of collective rights and responsibilities is also rooted in regional mechanisms such as the African Charter on Human and Peoples Rights, which calls for rights to be exercised within broader familial and common interest structures.[101] For instance, under an Indigenous Data Commons developed by the Canadian First Nations, Indigenous participation through stewardship is seen as critical to the OCAP (ownership, control, access and possession) of Indigenous data and therefore there is First Nation control over how information about them is collected, used and disclosed and on community access regardless of where the data is held.[102] In these frameworks, commons language anchors governance in sovereignty, relational accountability and collective rights, requiring Indigenous participation and authority across data life cycles.[103] There is currently guidance for researchers to meet Indigenous data sovereignty standards through reflective practices that foreground community values as well as advice on integrating Indigenous perspectives in research that may have direct or indirect effects on communities, such as the All of US health data collection initiatives.[104] From this perspective, some African scholars also make a compelling case for African data trusts to be rooted in collective principles such as ubuntu and ujamaa and Umoja, which call for data institutions to centre collective interests in ways that maintain equitable wealth distribution and broader community relations.[105] The CARE Principles discussed above explicitly highlight tensions between Indigenous rights and open data, machine learning, broad sharing and secondary use, and they treat benefit-sharing and authority to control as essential rather than aspirational.[106] OCAP similarly treats control over collection, use and disclosure as an expression of self-determination, and it foregrounds access rights and physical possession as mechanisms of protection.
H. Intersectionality
While data commons attempt to challenge power relations, there is also a risk that they may sustain them and create new ones. For instance, growing research on data extraction in the Global South shows that even where foreign control of data doesn’t exist, there is still extraction by an emerging middle class.[107] Speculative work on data commons can help surface these tensions and promote intersectional ethics grounded in collective rights when designing health data governance infrastructures. Feminist scholars Gurumurthy and Chami argue that confronting the ‘datafied regime of governmentality’, in the case of global markets for tracking products for women, requires shifting governance towards collective social control over the health data commons through a feminist, relational ethics. They propose community data stewardship bodies with public interest mandates that would be embedded in struggles over gender, caste and age.[108] The Aapti Institute India has also proposed a data stewardship playbook with similar values, which identifies structural inclusion as a core problem and suggests forging community-level partnerships that empower engaged community members to build bottom-up data-oriented communities while also facilitating more diverse onboarding for stewardship organisations.[109]
6. Conclusion
This article develops a theory of datafication as dispossession, from the starting point that health data is not simply taken from social life, as though it already exists in a ready-made form. Rather, health data is produced by abstracting social life into data. Clinical encounters, bodily vulnerability, genomic inheritance, public health work and community exposure are turned into informational objects that can be aggregated and interpreted and moved across institutions. This process creates value, but it also creates the conditions through which control can be lost. As health data passes through software systems, databases, AI models and proprietary infrastructures, authority over its meaning and benefits can shift away from the people, communities and health systems that made the data possible. The problem, then, is not only that value is extracted from health data, but that legal and technical systems mediate the transformation of human experience into something that can be accumulated by others.
The main proposals here interrupt dispossession at specific points. Legal frameworks should guarantee sustained benefit rights for communities in secondary uses of health data. Benefit rights for communities in secondary uses interrupt the moment when value is extracted without return, requiring that benefits from secondary uses such as research, products, innovations and services be shared back in defined, enforceable ways, for example, revenue shares, capacity building and local access to results. Legislation must grant local authorities audit powers to trace downstream uses of secondary data to interrupt the opacity of data flows, forcing documentation of derivative products, enabling remedies where terms are breached. Explainability requirements for high‑risk secondary uses interrupt the epistemic control exercised through uninterpretable models, mandating that systems expose their rationale, feature importance, and data sources at levels accessible to affected communities. Regulators and civil society organisations should be involved in impact assessments, and licensing and experimental regulation should involve quasi‑technical bodies with deep expertise. These bodies would assess licences, certify explainability standards, run sandboxes for experimental uses and advise on risk thresholds. They should combine technical, legal and community voices and operate with transparent procedures and public reporting. Invariably, this is going to involve many variations of data commons, but given that in many countries health is delivered in a decentralised way, this may help to layer bureaucracy in the unchecked power of amalgamating health data, and the localisation of health data commons will enable communities to remain in touch with their health data and rebuild the trust within global health systems that has been eroded in the past. Finally, refusal architecture – whether built into facility data stewards who block queries that fail template checks or into data stations that deny requests prioritising profit over the public health mission – interrupts the automatic flow of data by embedding a community veto power into the infrastructure itself. Accountability must rest with people and communities, not opaque systems. The rights to contest automated decisions, request model audits and demand remediation should be codified. Community representatives should have standing oversight of processes. Of course, this is made complex with health data because it may be technical, but the onus should be on data users to explain possible harms and on trained data stewards, who have explicit duties to represent data subjects by negotiating equitable benefits, monitoring terms and conditions and refusing extractive access.
In the Global South, this will likely be very complex, but capacity building projects in the commons are possible, and we have good exemplars, with the H3Africa and CIDACS projects, of overarching principles that will increase capacity in these areas. Some commons infrastructure, such as CIDACS, builds refusal into its architecture. Facility data stewards can block queries that fail template checks or violate local privacy rules or community agreements, and no data leaves the site unless all checks pass. PharmAccess’s blueprint similarly requires ‘stewarding access’, where data stations may deny requests that prioritise profit over the defined public health mission. A generation of data stewards from the Global South with both technical skills and the lived experience of delivering and receiving health care in post-colonial settings will be essential to reimagining and operating health data commons that protect local interests while enabling collective value.
Commons rhetoric can be mobilised to legitimate data pooling without resolving conflicts over value extraction and control, particularly when data infrastructures are built on platform-dependent architectures. In a neoliberal era, common ownership could also mean that collective rights are transferred with less accountability about extraction and use of health data. Thus, creating new data commons to respond to problems of colonial and inequitable access to data may simply make extraction liveable through these new legal forms that reinforce the same inequalities. Several legal principles are therefore important in the establishment of new health data commons. This points to a broader theoretical claim about intellectual property and the data form itself. The legal mechanisms of dispossession operate through IP regimes that recharacterise collective health knowledge as private property. Radical IP scholarship challenges whether IP should function as broad property rights constrained only by other property rights, or whether it should be recharacterised as limited rights explicitly oriented towards the public good.[110] This question is central to data commons. If IP remains a broad property regime, commons projects will be perpetually vulnerable to enclosure through licensing, contractual constraints and technical protection measures. The framework developed here treats the data form itself as a mode of dispossession, and IP is the primary legal technology through which that form becomes enforceable. The four proposals (benefit rights, audit powers, explainability and refusal architecture) are interventions that reconfigure the relationship between data form and the property regime, creating the conditions for health data commons that protect local interests while enabling collective value.
-
*Professor of Global Health Law, University of Warwick. Email: sharifah.sekalala@warwick.ac.uk. I would like to thank Shajoe Lake for research assistance as well as Tatenda Chatikobo, Ana Deborah De Silva and Yureshya Perrera for comments and conversations during the writing of this article. Thanks to all those who attended the UCL Current Legal Problems Lecture for their comments and questions, which helped me to make the article stronger. This work is supported by a Wellcome Trust Grant, ‘There is no app for this regulating the migration of digital health data in Sub Saharan Africa’ (Trust, grant number: 225238/Z/22/Z).
Akim Benamara on TechAfrica News Podcast, ‘Episode 7: Africa CDC’s Continental Vision – Health Tech for 1.4 Billion People’ (19 June 2025) <https://techafricanews.com/2025/06/19/episode-7-africa-cdcs-continental-vision-health-tech-for-1-4-billion-people/> accessed 18 June 2026. ↑
-
Caribbean Community, ‘CARICOM Secretary-General: AI, Genomics and Digital Health Will Transform Caribbean Public Health’ (CARICOM, 22 April 2026) <https://caricom.org/caricom-secretary-general-ai-genomics-and-digital-health-will-transform-caribbean-public-health/> accessed 18 June 2026 ↑
-
Daniel Doane, Ameera Kanjee and Jennifer Nelson, ‘Report Caribbean Connect: Building the Pan-American Highway for Digital Health’ (Inter-American Development Bank 2024) <https://doi.org/10.18235/0013204>. ↑
-
Luc Rocher, Julien M Hendrickx and Yves-Alexandre de Montjoye, ‘Estimating the Success of Re-Identifications in Incomplete Datasets Using Generative Models’ (2019) 10 Nature Communications 3069. ↑
-
Yu Wang and Fan Dong, ‘How Intellectual Property Protection Enhances Data Capitalization: The Mediating Effects of Resources and Willingness to Capitalize on Data’ (2025) 107 International Review of Financial Analysis 104539. ↑
-
Sylvie Delacroix and Neil D Lawrence, ‘Bottom-up Data Trusts: Disturbing the “One Size Fits All” Approach to Data Governance’ [2019] International Data Privacy Law ipz014. ↑
-
Datafication refers to the use of new communications and database technologies to generate digital data that is machine-readable and computationally manipulable, especially for big data analytics. See Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution That Will Transform How We Live, Work, and Think (Houghton Mifflin Harcourt 2013) cited in Linnet Taylor and Dennis Broeders, ‘In the Name of Development: Power, Profit and the Datafication of the Global South’ (2015) 64 Geoforum 229, 229. ↑
-
Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (Profile Books 2019). ↑
-
See eg Nick Couldry and Ulises A Mejias, The Costs of Connection: How Data Is Colonizing Human Life and Appropriating It for Capitalism (Stanford University Press 2019); Michael Kwet, ‘Digital Colonialism: US Empire and the New Imperialism in the Global South’ (2019) 60 Race & Class 3; Sharifah Sekalala and Tatenda Chatikobo, ‘Colonialism in the New Digital Health Agenda’ (2024) 9 BMJ Global Health e014131; Toussaint Nothias, ‘An Intellectual History of Digital Colonialism’ [2025] Journal of Communication jqaf003. ↑
-
Sara Rosenbaum, ‘Data Governance and Stewardship: Designing Data Stewardship Entities and Advancing Data Access’ (2010) 45 Health Services Research 1442; Delacroix and Lawrence (n 6); Marina Micheli and others, ‘Emerging Models of Data Governance in the Age of Datafication’ (2020) 7 Big Data & Society 2; Cynthia Srnec and Alexandre Guttmann, ‘Platform Cooperatives, a Model of Commons and Sustainability’ in Jerome Nikolai Warren and others (eds), The Routledge Handbook of Cooperative Economics and Management (Routledge 2024); Mark Esposito, ‘Cooperatives and the Digital Commons: Governance, Sustainability, and Shared Infrastructure’ (Platform Cooperativism Consortium, 18 March 2025) <https://platform.coop/blog/cooperatives-and-the-digital-commons-governance-sustainability-and-shared-infrastructure/> accessed 18 June 2026. ↑
-
Elinor Ostrom, Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge University Press 1990). ↑
-
S Ciriacy-Wantrup and Richard Bishop, ‘Common Property as a Concept in Natural Resources Policy’ (1975) 15 Natural Resources Journal 713. ↑
-
David Feeny and others, ‘The Tragedy of the Commons: Twenty-Two Years Later’ (1990) 18 Human Ecology 1. ↑
-
David Harvey, ‘Accumulation by Dispossession’ in David Harvey, The New Imperialism (Oxford University Press 2003); Mark Neocleous, ‘International Law as Primitive Accumulation; Or, the Secret of Systematic Colonization’ (2012) 23 European Journal of International Law 941; Katharina Pistor, The Code of Capital: How the Law Creates Wealth and Inequality (Princeton University Press 2020). ↑
-
Aaron Windel, Cooperative Rule: Community Development in Britains Late Empire (University of California Press 2022). ↑
-
Clemens Scott Kruse and others, ‘The Use of Electronic Health Records to Support Population Health: A Systematic Review of the Literature’ (2018) 42 Journal of Medical Systems 214. ↑
-
Rocher, Hendrickx and de Montjoye (n 4); Victor Janmey and Peter L Elkin, ‘Re-Identification Risk in HIPAA De-Identified Datasets: The MVA Attack’ [2018] 2018 AMIA Annual Symposium proceedings 1329. ↑
-
Wang and Dong (n 5). ↑
-
Sharifah Sekalala and others, ‘America First, Africa Last? Health Data Deals and the New Scramble for Pathogens’ (2026) 6 PLOS Global Public Health e0005974. ↑
-
Mirambo v Independent Electoral and Boundaries Commission; Smartmatic International Holdings (Interested Parties) (Constitutional Petition E488 & E465 of 2021 (Consolidated)) [2022] KEHC 11763 (KLR) (Constitutional and Human Rights) (16 May 2022) (Judgment). Also see Africa Centre for Open Governance and Kenyans for Peace with Truth and Justice, Disobeying Orders: The Scrutiny of Kenya’s Electoral Technology, ICT Experts’ Report for Presidential Petition No E003 of 2022 on the Scrutiny of IEBC’s Electoral Technology (AfriCOG and KPTJ 2022) <https://africog.org/wp-content/uploads/2022/12/ICT_Experts_Report_For_Kenya_Presidential_Election_Petition_2022.pdf> accessed 18 June 2026. ↑
-
‘Hopu Kōrero Māori Group Competition’ (Te Hiku Media, 9 March 2018) <https://tehiku.nz/p.BnC> accessed 18 June 2026 . ↑
-
Grimshaw Club, ‘The Use of Māori Knowledge in AI: How Does It Affect Indigenous Data Sovereignty and Human Rights?’ (Grimshaw Club, 28 February 2025) <https://www.grimshawclub.org/post/the-use-of-māori-knowledge-in-ai-how-does-it-affect-indigenous-data-sovereignty-and-human-rights> accessed 18 June 2026. ↑
-
Segun Adeyemi, ‘Why Africans Don’t Trust Bill Gates’ $50 Million Health Initiative in Africa’ (Business Insider Africa, 22 January 2026) <https://africa.businessinsider.com/local/markets/why-africans-dont-trust-bill-gates-dollar50-million-health-initiative-in-africa/9fgs9hz> accessed 18 June 2026. ↑
-
Nothias (n 9). ↑
-
Couldry and Mejias (n 9). ↑
-
Kwet (n 9). ↑
-
Abeba Birhane, ‘Algorithmic Colonization of Africa’ (2020) 17 SCRIPTed 389 <https://doi.org/10.2966/scrip.170220.389>. See also Mejias and Couldry, Data Grab (Penguin 2024). ↑
-
Sharifah Sekalala, ‘Ending Pandemics within the Shadow of Trade: Reconciling Equity in Global Health with Coloniality’ (2025) 14 Cambridge International Law Journal 4 <https://doi.org/10.4337/cilj.2025.01.01>. ↑
-
ibid. ↑
-
Donato Paolo Mancini, ‘Dispute over Genomic Databank Undermines Fight to Thwart next Pathogen’ Financial Times (18 March 2024). ↑
-
Salome Viljoen, ‘Democratic Data: A Relational Theory For Data Governance’ [2020] SSRN Electronic Journal <https://doi.org/10.2139/ssrn.3727562>. ↑
-
Elettra Bietti, ‘Consent as a Free Pass: Platform Power and the Limits of the Informational Turn’ (2020) 40 Pace Law Review 310; Daniel Solove, ‘Murky Consent: An approach to the Fictions of Consent in Privacy Law’ (2024) 104 Boston University Law Review 593 <https://dx.doi.org/10.2139/ssrn.4333743>. ↑
-
WHO Regional Office for Europe, Health Data Governance in the Age of Artificial Intelligence: Policy Imperatives for the WHO European Region (2025). ↑
-
Nick Couldry and Ulises A Mejias, ‘Data Colonialism: Rethinking Big Data’s Relation to the Contemporary Subject’ (2019) 20 Television & New Media 336. ↑
-
National Cancer Institute at the National Institutes of Health, ‘Data Commons’ <https://datacommons.cancer.gov/explore/data-commons> accessed 18 June 2026. ↑
-
Michael Cox and others, ‘Synthesizing Theories of Natural Resource Management and Governance’ (2016) 39 Global Environmental Change 45. ↑
-
Amineh Ghorbani and Giangiacomo Bravo, ‘Managing the Commons: A Simple Model of the Emergence of Institutions through Collective Action’ (2016) 10 International Journal of the Commons 200. ↑
-
Katherine J Strandburg, Brett M Frischmann and Michael J Madison, ‘The Knowledge Commons Framework’ in Brett M Frischmann, Katherine J Strandburg and Michael J Madison (eds), Governing Medical Knowledge Commons (Cambridge University Press 2014). ↑
-
Charlotte Hess and Elinor Ostrom, Understanding Knowledge as a Commons: From Theory to Practice (MIT Press 2007). ↑
-
Strandburg, Frischmann and Madison (n 38). ↑
-
Nadya Purtova and Gijs Van Maanen, ‘Data as an Economic Good, Data as a Commons, and Data Governance’ (2024) 16 Law, Innovation and Technology 1 <https://doi.org/10.1080/17579961.2023.2265270>. ↑
-
Alek Tarkowski and Jan J Zygmuntowski, ‘Data Commons Primer. Democratizing the Information Society’ (Open Future, August 2022) 8 <https://openfuture.eu/wp-content/uploads/2022/07/220723data-commons-primer.pdf> accessed 18 June 2026. ↑
-
Sylvie Delacroix, ‘Sustainable Data Rivers? Rebalancing the Data Ecosystem That Underlies Generative AI’ (2024) 2 Critical AI <https://doi.org/10.1215/2834703X-11205224>. ↑
-
Giuliano G Castellano, Ēriks K Selga and Douglas W Arner, ‘The Emergence of Financial Data Governance and the Challenge of Financial Data Sovereignty’ in Anupam Chander and Haochen Sun (eds), Data Sovereignty: From the Digital Silk Road to the Return of the State (Oxford University Press 2023) 178. ↑
-
See Walter D Mignolo, ‘Epistemic Disobedience, Independent Thought and Decolonial Freedom’ (2009) 26 Theory, Culture & Society 159; Walter D Mignolo, The Darker Side of Western Modernity: Global Futures, Decolonial Options (Duke University Press 2011); Aníbal Quijano, Aníbal Quijano: Foundational Essays on the Coloniality of Power (Duke University Press 2024). ↑
-
Purtova and Van Maanen (n 41). ↑
-
Brett Frischmann, ‘An Economic Theory of Infrastructure and Commons Management’ (2005) 89 Minnesota Law Review 917; Joan Rodon Mòdol, ‘Citizens’ Cooperation in the Reuse of Their Personal Data: The Case of Data Cooperatives in Healthcare’ [2019] Progress in IS 159, 161–62 <https://ideas.repec.org//h/spr/prochp/978-3-319-94487-6_8.html> accessed 18 June 2026. ↑
-
Mòdol (n 47). ↑
-
Luc Rocher and Jessica Morley, ‘Protecting Health Data at UK Biobank’ (2026) 393 British Medical Journal 660. ↑
-
Strandburg, Frischmann and Madison (n 38). ↑
-
Hess and Ostrom (n 39). ↑
-
James Boyle, ‘The Second Enclosure Movement and the Construction of the Public Domain’ (2003) 66 Law and Contemporary Problems 33. ↑
-
ibid. ↑
-
ibid. ↑
-
Purtova and Van Maanen (n 41); Nadezhda Purtova, ‘Health Data for Common Good: Defining the Boundaries and Social Dilemmas of Data Commons’ in Samantha Adams, Nadezhda Purtova and Ronald Leenes (eds), Under Observation: The Interplay Between eHealth and Surveillance (Springer 2017). ↑
-
Katharina Pistor, The Law of Capitalism and how to Transform it (Yale University Press 2025). ↑
-
UNESCO, ‘DHIS2 – UNESCO Digital Innovation Hub’ <https://www.unesco.org/en/unesco-digital-innovation-hub/dhis2> accessed 18 June 2026. ↑
-
Jørn Braa and Sundeep Sahay, ‘The DHIS2 Open Source Software Platform: Evolution Over Time and Space’ in Leo Anthony G Celi and others (eds), Global Health Informatics. Principles of eHealth and mHealth to Improve Quality of Care (MIT Press 2017) 451. ↑
-
Robert L Grossman, ‘Data Lakes, Clouds, and Commons: A Review of Platforms for Analyzing and Sharing Genomic Data’ (2019) 35 Trends in Genetics 223. ↑
-
Vili Lehdonvirta, Cloud Empires: How Digital Platforms Are Overtaking the State and How We Can Regain Control (MIT Press 2022). ↑
-
Fiona McCormack, ‘Māori Saltwater Commons: Property, Wealth, and Inequality’ (2018) 1 Commoning Ethnography 9. ↑
-
Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution That Will Transform How We Live, Work, and Think (John Murray 2013) 1–18. ↑
-
Jennifer Shkabatur, ‘The Global Commons of Data’ (2019) 22 Stanford Technology Law Review 1. ↑
-
Delacroix and Lawrence (n 6). ↑
-
Caroline AB Redhead and others, ‘Unlocking the Promise of UK Health Data: Considering the Case for a Charitable GP Data Trust’ (2025) 33 Medical Law Review 1. ↑
-
Rosenbaum (n 10); Benjamin Bartlett and others, ‘Health Data Stewardship: Achieving Trust through Accountability in Health Data Sharing for Research’ (2024) 16 Law, Innovation and Technology 517 <https://doi.org/10.1080/17579961.2024.2392937>. ↑
-
Micheli and others (n 10). ↑
-
Rosenbaum (n 10). ↑
-
Viljoen (n 31); Roshni Modhvadia and Octavia Field Reid, ‘Participatory and Inclusive Data Stewardship: A Landscape Review’ (Ada Lovelace Institute, 13 December 2024) <https://www.adalovelaceinstitute.org/report/participatory-inclusive-data-stewardship/> accessed 18 June 2026. ↑
-
Open Data Institute, ‘What Are Data Institutions and Why Are They Important?’ (29 January 2021) <https://theodi.org/insights/explainers/what-are-data-institutions-and-why-are-they-important/> accessed 18 June 2026. ↑
-
Rosenbaum (n 10). ↑
-
Ada Lovelace Institute, Exploring Legal Mechanisms for Data Stewardship (2021) <https://www.adalovelaceinstitute.org/wp-content/uploads/2021/03/Legal-mechanisms-for-datastewardship_report_Ada_AI-Council-2.pdf> accessed 18 June 2026 ↑
-
Sylvie Delacroix, ‘“Data Steward” is Just One of the Missing 21st-Century Professions That Universities Could be Creating’ (Wonkhe Blog, 5 February 2026) <https://wonkhe.com/blogs/data-steward-is-just-one-of-the-missing-21st-century-professions-that-universities-could-be-creating/> accessed 18 June 2026; Modhvadia and Reid (n 69). ↑
-
Teanna Barrett and others, ‘African Data Ethics: A Discursive Framework for Black Decolonial Data Science’ (arXiv, 17 June 2025) <https://doi.org/10.48550/arXiv.2502.16043>. ↑
-
Chih-hsing Ho and Tyng-Ruey Chuang, ‘Governance of Communal Data Sharing’ in Angela Daly, S Kate Devitt and Monique Mann (eds), Good Data (Institute of Network Cultures 2019) 202. ↑
-
Involve, ‘Public Data for Public Good: A Report by Involve for the Liverpool Civic Data Co-operative’ <https://civicdatacooperative.com/app/uploads/2022/11/Publicdataforpublicgoodreport.pdf> accessed 18 June 2026. ↑
-
Covenant of the League of Nations (adopted 28 April 1919, entered into force 10 January 1920) art 22. ↑
-
Joseph Kieyah, ‘Indigenous Peoples’ Land Rights in Kenya: A Case Study of the Maasai and Ogiek People’ (2007) 15 Penn State Environmental Law Review 397. ↑
-
Tony Bennett, The Birth of the Museum: History, Theory, Politics (Reprinted, Routledge 2009); Dan Hicks, The Brutish Museums: The Benin Bronzes, Colonial Violence and Cultural Restitution (Paperback edn, Pluto Press 2021). ↑
-
Hicks (n 79). ↑
-
See Silvia Federici, Re-Enchanting the World: Feminism and the Politics of the Commons (PM Press 2019). ↑
-
Daniel Kapitan and others, ‘Towards a Health Data Commons in LMICs: Demonstrating Fair Sharing and Reuse of Health Data in Sub-Saharan Africa’ (PharmAccess) <https://health-data-commons.pharmaccess.org//Towards-a-health-data-commons-in-LMICs.pdf> accessed 18 June 2026. ↑
-
ibid 29. ↑
-
ibid 29–33. ↑
-
Mirjam van Reisen and others, ‘Design of a FAIR Digital Data Health Infrastructure in Africa for COVID‐19 Reporting and Research’ (2021) 2 Advanced Genetics e10050 <https://doi.org/10.1002/ggn2.10050>. ↑
-
Kapitan and others (n 82) 21. ↑
-
ibid 24. ↑
-
Tatenda Chatikobo and Sharifah Sekalala, ‘A Critique of the Emancipatory Promise of Open-Source Software in Digital Health in LMICs’ (2026) 8 Law, Technology and Humans 44 <https://doi.org/10.5204/lthj.4165>. ↑
-
John K Waters, ‘Google’s Data Commons Launches Server to Connect AI Systems with Public Datasets’ (Pure AI, 26 September 2025) <https://pureai.com/articles/2025/09/26/google-data-commons-launches-server-to-connect-ai-systems-with-public-datasets.aspx.> accessed 18 June 2026. ↑
-
RV Guha and others, ‘Data Commons’ (Google, 12 September 2023) <https://storage.googleapis.com/gweb-research2023-media/pubtools/7312.pdf> accessed 18 June 2026. ↑
-
Couldry and Mejias (n 9). ↑
-
Waters (n 89). ↑
-
Erika Check Hayden, ‘Proposed Ebola Biobank Would Strengthen African Science’ (2015) 524 Nature 146 <https://doi.org/10.1038/524146a>. ↑
-
Stephanie Russo Carroll and others, ‘The CARE Principles for Indigenous Data Governance’ (2020) 19 Data Science Journal 1 <https://doi.org/10.5334/dsj-2020-043>. ↑
-
ibid. ↑
-
Edith Brown Weiss, ‘Our Rights and Obligations to Future Generations for the Environment’ (1990) 84 American Journal of International Law 198 <https://doi.org/10.2307/2203020>; Caroline E Foster, ‘Due Regard for Future Generations? The No Harm Rule and Sovereignty in the Advisory Opinions on Climate Change’ (2024) 13 Transnational Environmental Law 588 <https://doi.org/10.1017/s2047102524000207>. ↑
-
H3Africa Biorepository Program, ‘H3Africa Biorepository Program’ <https://www.biorepository.africa/> accessed 18 June 2026. ↑
-
ML Barreto and others, ‘The Centre for Data and Knowledge Integration for Health (CIDACS): Linking Health and Social Data in Brazil’ (2019) 4 International Journal of Population Data Science 1140. ↑
-
Bethania de Araujo Almeida and others, ‘CIDACS’ Efforts towards an Inclusive and Dialogic Data Governance in Brazil: A Focused Literature Review’ (2024) 9 International Journal of Population Data Science 2163. ↑
-
Daniela Almeida and others, ‘Examining the Quality of Record Linkage Process Using Nationwide Brazilian Administrative Databases to Build a Large Birth Cohort’ (2020) 20 BMC Medical Informatics and Decision Making 173. ↑
-
African Charter on Human and Peoples’ Rights (adopted 27 June 1981, entered into force 21 October 1986) 1520 UNTS 217, arts 27(2) and 29. ↑
-
First Nations Information Governance Centre, ‘The First Nations Principles of OCAP®’ <https://fnigc.ca/ocap-training/> accessed 18 June 2026. ↑
-
Carroll and others (n 94). ↑
-
Kalinda E Griffiths and others, ‘Indigenous and Tribal Peoples Data Governance in Health Research: A Systematic Review’ (2021) 18 International Journal of Environmental Research and Public Health 10318; Theresa Patten and others, ‘The All of Us Research Program’s Wearables Dataset’ [2026] Nature Medicine 1. ↑
-
Talitha Hlaka, ‘Data Governance in the African Context’ (MERL Tech, 5 April 2022) <https://merltech.org/data-governance-in-the-african-context/> accessed 18 June 2026; Nokuthula Olorunju and Rachel Adams, ‘African Data Trusts: New Tools towards Collective Data Governance?’ (2024) 33 Information & Communications Technology Law 85 <https://doi.org/10.1080/13600834.2023.2260678>. ↑
-
Carroll and others (n 94). ↑
-
Sharifah Sekalala and Tatenda Chatikobo, ‘Digital Health for Shared Value: A Critique of Legal Infrastructures in a Post-Colonial Context’ (2026) 22 International Journal of Law in Context 11. ↑
-
Anita Gurumurthy and Nandini Chami, ‘Beyond Data Bodies: New Directions for a Feminist Theory of Data Sovereignty’ [2023] SSRN Electronic Journal <https://doi.org/10.2139/ssrn.4037321>. ↑
-
Modhvadia and Reid (n 70) 72. ↑
-
Delacroix (n 43); Alexander Peukert, A Critique of the Ontology of Intellectual Property Law (Gill Mertens tr, Cambridge University Press 2021). ↑