The Canadian government announced funding for a national health-sector data space in June 2026. The details are still to come, but data spaces typically consist of a shared set of rules, standards and infrastructures that enable data to be used across organizations and borders.
Canada is not the first to try this. The European Union spent years negotiating its own health data space, which covers 17 categories of health data and 27 EU member states. In March 2025, the EU’s Regulation on the European Health Data Space came into force.
Canada’s health-sector data space announcement was part of the renewed national AI strategy, following the deputy ministers of health’s 2023 endorsement of the Pan-Canadian Health Data Charter.
Canada has the tech to connect health records. Why isn’t it being used?
Canada’s health data is flowing abroad while Ottawa stalls on AI rules
AI scribes in health care raise risks for patients and privacy
AI requires data, and the health sector is a priority area for AI development and implementation. Growing concern about digital sovereignty — as Prime Minister Mark Carney set out at Davos — is reinforcing the federal government’s commitment to invest in secure infrastructure for sensitive data.
Infrastructure of this scale is never neutral. Decisions about how data spaces are built are consequential; they determine whose interests are served and who does or does not benefit from them. The EU’s experience developing the European Health Data Space, or EHDS, offers a preview of the choices Canada will face, as well as some important lessons to consider.
Institutionalize the longer-term goals
Canada’s health systems face challenges such as equitable access and workforce sustainability. While these challenges have complex causes, data and AI are increasingly positioned as solutions to improve productivity and service delivery.
Canada and the EU offer contrasting approaches to realizing this potential. Canada’s health data charter outlines principles for health data design (for example, people-centred, co-operative, accessible, innovative) with future value tied to practices that align with those principles. The expectation is that systems designed according to these principles will generate value for patients, providers, researchers and health systems.
Rather than beginning with broad principles, the EHDS establishes rights and rules governing how health data is accessed, shared and re-used by member states. It guarantees patient access to electronic health records free of charge and explicitly ensures the availability of health data for research, innovation and policymaking purposes. In doing so, the data space embeds expectations about the public value of health data in institutional arrangements determining permitted uses and minimum data categories.
While each of these approaches has merit, reliance on principles, as in Canada’s system, can go only so far; principles require translation to practices to achieve change. Moving to practices from principles will require articulating what we expect AI and data to do and for whom.
The European data space has this partly right: it is a regulation that dictates purpose, data users and data categories. However, it remains a creature of the EU’s broader data strategy in which value is assumed to follow from European competitiveness and sovereignty. Canada is poised to repeat this move. The Canadian data space is advancing under the umbrella of a national AI strategy with the assumption that AI adoption will generate public value through job creation, efficiencies and economic growth.
Canada must reverse this approach: rather than assuming public value follows from technological advancement, value definition must incorporate diverse perspectives, including members of the public, as a precondition for how health data infrastructures are designed, governed and evaluated.
Accommodate differences thoughtfully
The EHDS represents a massive effort to scale health data access and use from single countries to all 27 EU member states and, eventually, third countries such as Canada. Canada faces a parallel task in scaling from individual provinces and territories to the country as a whole. Canada also has the distinct needs of a constitutional federation and an obligation to support Indigenous-data sovereignty in line with the United Nations Declaration on the Rights of Indigenous Peoples framework and related legislation at the federal level as well as in some provinces.
Scaling AI and the data infrastructures that support it requires developing governance practices capable of addressing differences between jurisdictions with distinct responsibilities, communities with different interests and rights, and competing expectations about how health data should generate value.
The EU’s health data space establishes a baseline for data access and re-use for secondary purposes, including AI model training. The EHDS specifies who may access data, which purposes are permitted and which prohibited, and how individuals may exercise certain rights. However, it leaves fees, composition of health data access bodies and additional safeguards as decisions for member states. These are deliberate institutional design choices about how to co-ordinate across different health systems, legal traditions and publics. They are not inherent features of the infrastructures themselves.
Early implementation in the EU suggests both the promise and the risk of this approach: flexible member-state participation, but diverging choices on opt-outs and fees threaten the consistency the EHDS was meant to deliver. Canada should set its common floor deliberately, with an eye to the unintended consequences of those choices.
Build both social and technical infrastructure
Work on the data spaces in both the EU and Canada requires investment, not only in technical infrastructure, but also in the social processes and capabilities that make it possible to deliver value at scale for patients, health systems, developers and data holders.
Strong Canadian-based technical infrastructure can support sovereignty, but choices need to be made about who should and should not control it and who should benefit from it. These choices will influence whether what we build will be trusted, fair and worth having. Making good decisions will rely on complementary capabilities to evaluate data-intensive innovation, including new approaches to responsible design, evaluation, cost-benefit assessments, accountability and public engagement.
These capabilities and activities should not be viewed as separate from infrastructure or as an afterthought to it. They are the conditions under which infrastructure earns trust and delivers real value for people in Canada. Meaningful involvement – including by members of the public – in health data and AI governance would be consistent with the Pan-Canadian Health Data Charter. Evidence-based approaches to addressing declining public trust will be important.
This is where the European data space struggled most during its development. Concerns of member states, data holders and users about secondary use forced significant concessions at the last minute, including a general right to opt out. Canada has the opportunity to treat engagement as a critical design input at the outset.
Don’t mistake movement for progress
Understanding the use of health data and AI as questions of value-oriented public governance – answerable in law and revisable through deliberation – is at odds with treating implementation and scale as technical details to be settled after systems are already running. The question is not only how to leverage AI and health data spaces, but also what kinds of health systems we want and how building a health-sector data space can help get us there.
The capabilities of AI are changing rapidly. Its adoption in health care is remarkably fast for a sector known for moving slowly from innovation to implementation. But health systems become resilient and responsive only with vision and direction. Change without that is drift – motion we mistake for progress because it is fast.
