TORONTO — Medical researchers and legal experts emphasize the urgent necessity to protect a valuable Canadian resource: patient health information that could be utilized to train artificial intelligence (AI). Natalie Raffoul, an intellectual property lawyer based in Ottawa, asserts that Canada's health data represents one of the most valuable health datasets globally. She points out that no other jurisdiction can gather such a comprehensive data set due to Canada's unique public health system and its ethnic diversity.
Experts highlight that many Canadian institutions currently rely on cloud servers operated by American companies for storing health data. This situation, combined with former President Donald Trump’s ambition to position the U.S. as a leader in AI and speculation about his desire to make Canada a 51st state, raises concerns about the potential for U.S. government access to this data, possibly citing national security as a justification.
Dr. Amol Verma, a professor specializing in AI research and medical education at the University of Toronto, explains that algorithms used in healthcare require training on highly representative data to yield accurate results. He notes that U.S. health data lacks the inclusivity present in Canadian data, primarily due to the private healthcare system in the U.S., which often leaves uninsured individuals without accessible care and, thus, unrecorded health information. Consequently, AI models trained exclusively on U.S. data may exhibit biases concerning certain racial or linguistic groups.
Dr. Kumanan Wilson, research chair in digital health innovation at the University of Ottawa, underscores the potential economic value of Canadian health information to the U.S. He points out that major cloud providers, including Amazon Web Services (AWS), Microsoft Azure, and Google Cloud, could be legally compelled to grant access to health data stored by Canadian hospitals if the Trump administration sought it.
Michael Geist, Canada Research Chair in internet and e-commerce law at the University of Ottawa, expresses that while data privacy concerns were not prominent prior to Trump’s election, recent political developments necessitate a thorough reevaluation of data protection strategies. In response to inquiries from The Canadian Press, both Microsoft and Google refrained from commenting on hypothetical scenarios regarding access to data but indicated that any governmental requests would require a lawful court order or warrant. AWS also confirmed that they would only disclose data if compelled by a legally enforceable order.
Despite assurances from these companies regarding their data protection practices, Raffoul and Geist caution that storing Canadian health data, even within Canadian borders, might create a false sense of security if the service provider is American. Raffoul emphasizes that putting sensitive data into the hands of a Canadian subsidiary of a foreign multinational amounts to relinquishing it to foreign control. Geist expresses skepticism about the reliability of privacy measures under the Trump administration, stating that Canadian privacy laws lack robust penalties, which has been a persistent criticism. He suggests the introduction of a "blocking statute" that would impose severe penalties on U.S. companies for disclosing Canadian data, creating a legal rationale against compliance with U.S. orders.
Experts unanimously advocate for the establishment of more Canadian-owned tech companies dedicated to health data storage as the most effective solution to this dilemma. Additionally, they acknowledge Canada’s potential as a global leader in AI, with noteworthy contributions from figures such as Geoffrey Hinton, often referred to as the "Godfather of AI," who has conducted much of his groundbreaking work at the University of Toronto and the Vector Institute.
Verma points out that the current wave of patriotism in Canada, incited by U.S. political posturing, presents a unique opportunity to consolidate health data across provincial and territorial lines, potentially enhancing AI algorithms in healthcare. He emphasizes that this moment could allow Canada to leverage its strengths to lead in the AI domain, but warns that complacency could result in falling behind.
This report was first published on April 24, 2025, by The Canadian Press, which benefits from a partnership with the Canadian Medical Association.
