Empowering teams with clarity, consistency, and lasting impact across sectors.

‹ Back

Canada is building the health data infrastructure for AI. Is governance keeping up?

Canada is making a significant move toward building the infrastructure needed to scale artificial intelligence in health care. On June 10, 2026, the Canadian Institute for Health Information (CIHI) announced that it will work with governments and health data organizations to create a Health Sector Data Space, supported by a $100 million investment announced as part of Canada’s national AI strategy, AI for All. The initiative is intended to connect secure, private and standardized health datasets to strengthen clinical research, health services research and health system performance measurement.

Canada has enormous amounts of health data, but much of it remains fragmented across provinces, territories, institutions and systems, making a sector specific data space ambitious, and important. The Health Sector Data Space is intended to create better ways of connecting and using assets while allowing organizations to continue stewarding and controlling their own data. For AI, that matters enormously as it is only as useful as the data, infrastructure and governance surrounding it.

This is where Canada's opportunity becomes a governance challenge because the data problem was never simply a data problem. For years, conversations about AI in health care have focused on whether Canada has enough data to compete, and it is clear, we do. The more difficult question is whether we can connect, standardize, govern and use that data responsibly at scale.

A national health data space will create enormous opportunities: better clinical research, more sophisticated health system planning, improved measurement of outcomes, earlier identification of system pressures and AI models that better reflect Canadian patients and healthcare environments. CIHI is already exploring this direction, including partnerships aimed at developing sovereign AI models trained on Canadian health data, making this a significant step. However, building infrastructure that makes data more accessible does not automatically create the conditions for responsible AI. In fact, it may expose a much larger gap.

I have been increasingly interested in the AI innovation governance gap: the widening distance between what technology allows organizations to do and what their governance systems are prepared to oversee. Health care makes this gap particularly consequential as

AI is already moving into clinical documentation, diagnostics, medical imaging, patient triage, scheduling, resource allocation, predictive analytics and decision support. As access to high-quality Canadian health data improves, the number and sophistication of these applications will only increase. The question therefore cannot simply be: Can we build it?; it must also be: Who is accountable when we do?

A health data ecosystem capable of supporting AI needs more than privacy protections and technical security. It needs governance across the entire AI lifecycle, from data collection and model development through validation, procurement, deployment, monitoring and eventual retirement. That means knowing what data was used, who controls it, whose experiences are represented, how a model was validated and whether it performs equitably across populations and care environments. It also means defining who is accountable when an AI supported decision is wrong, how clinicians can challenge its outputs, and who monitors for bias, model drift and unintended consequences.

These are governance questions, not technology questions and they need to be answered before AI becomes embedded in critical healthcare workflows. One concept deserves much more attention in the conversation about healthcare AI: the human control point. The objective should not be to keep humans nominally "in the loop" while designing systems that effectively make decisions for them. Meaningful human oversight requires a defined point at which a qualified person has the authority, information and capacity to question, override or stop an AI-supported decision. 

That distinction becomes increasingly important as AI moves from administrative efficiency toward clinical and system level decision making. An AI tool that drafts a clinical note is fundamentally different from one that influences a diagnosis. A system that identifies patients who may require additional follow up is different from one that determines who receives scarce resources. The greater the potential impact on a person, the stronger the governance surrounding the AI should be. Therefore, innovation should not determine the level of oversight, risk should.

The Health Sector Data Space provides Canada with an opportunity to do something much more ambitious than simply connect datasets., it can become a foundation for trusted AI infrastructure. That means embedding governance into the architecture rather than treating it as an approval process at the end. CIHI's focus on ethical, transparent and trustworthy AI is encouraging, but the next challenge is translating those principles into operational mechanisms.

Organizations need governance frameworks that can answer the following questions consistently and transparently:

• Is this AI appropriate for this use case?
• What evidence is sufficient before deployment?
• What level of human oversight is required?
• What monitoring must continue after implementation?
• When should an organization stop using an AI system?

These questions should become as routine as privacy impact assessments, clinical safety reviews and cybersecurity assessments.

Health care may be one of the prime sectors where the innovation-governance gap matters most, and where the consequences of getting it wrong are greatest. AI can help address some of Canada's most persistent health system challenges, however accelerating adoption without accelerating governance risks creating systems that can make or influence decisions faster than institutions can understand, challenging or governing them.

The answer is not to slow innovation, but more so to build governance at the speed of innovation. Canada now has an opportunity to do exactly that through the Health Sector Data Space, which can provide the infrastructure to connect Canada's health data. The next layer must be the infrastructure of trust: clear accountability, transparent decision making, responsible data stewardship, explainability, continuous monitoring and meaningful human control.

The real measure of success will not be how much health data Canada can connect. It will be whether we can turn that data into innovation without losing accountability along the way. This matters because in health care, the question is no longer whether AI is coming, it is whether our governance systems will be ready when it starts to produce data.

CT