OTTAWA — Canadians who follow AI health advice tools are five times more likely to experience harm than those who do not, according to a new survey from the Canadian Medical Association that highlights growing concerns about the intersection of AI technology and healthcare access.
The CMA released findings Tuesday from its third Health and Media Tracking Survey, conducted by Abacus Data, revealing a troubling pattern: while only 27% of Canadians trust AI to provide accurate health information, approximately half are using AI tools to diagnose or treat their health issues.
The survey, which polled 5,000 Canadians between November 3-13, 2025, paints a picture of a population caught between limited healthcare access and the readily available but potentially unreliable information provided by AI platforms.
The Access Crisis Driving AI Adoption
The survey found that most Canadians (89%) go online for health information for a variety of reasons, including that it’s faster and more convenient than trying to access care through the health system.
An estimated 5.9 million Canadians don’t have regular access to a primary care provider, down from 6.5 million in 2022, according to the OurCare Survey 2025 conducted by Dr. Tara Kiran in partnership with the CMA. However, access challenges persist: only 26% of Canadians obtained same or next-day appointments in 2023, according to data from the Canadian Institute for Health Information.
“For years, we’ve been talking about how too many Canadians struggle to access health care when they need it,” says Dr. Margot Burnell, CMA president. “This leaves people little choice but to turn to dubious sources of information, and now we know that it is hurting them. We fear what this will mean for patients in the future if action is not taken now.”
The survey revealed that 57% of respondents only turned to the internet when they couldn’t access a family doctor or other health care professional, according to reporting by The Globe and Mail, underscoring how healthcare system limitations are driving the shift to AI consultations.
The AI Accuracy Problem
Research has raised significant questions about AI’s reliability for medical information. Studies have shown that AI platforms like ChatGPT demonstrate variable accuracy when addressing health queries.
A study examining 150 medical case studies found that ChatGPT provided a correct diagnosis only 49% of the time, according to research reported by Live Science. A systematic review published in the journal Healthcare in January 2025 examining ChatGPT’s performance across medical specialties found an overall mean accuracy of 73.4%, though this varied significantly across different fields.
A recent study published in The Lancet Digital Health in February 2026 found that smaller or less advanced AI models believed false medical claims more than 60% of the time, while stronger systems like ChatGPT-4o did so in only 10% of cases.
Dr. Burnell explained that AI information presents unique challenges compared to traditional online health research. “AI information is different because it’s hard to know who’s generated it, it’s hard to ensure where the validity and the science has come from,” she told The Canadian Press.
Eroding Trust in Health Information
The CMA survey also revealed broader concerns about health information quality:
- 77% of Canadians are concerned about an increase in false health information coming from the United States
- 69% have become skeptical of any health information they find online, even from sources they think they should trust
- Despite these concerns, 85% of Canadians trust physicians to help them navigate health information
Nearly all survey respondents believe responsibility for curbing false health information falls to social media platforms (87%) and governments (90%).
Looking Forward
The CMA submitted recommendations to Innovation, Science and Economic Development Canada in October 2025 aimed at shaping AI development in ways that strengthen trust, protect privacy, and enhance healthcare.
“AI is already reshaping the delivery of health services, and it’s not going away,” says Dr. Burnell. “It is critical that government, health providers, patients work together to ensure AI is a tool for reliable health information.”
The association is calling on the federal government to take swift action to protect Canadians from false health information, invest in domestic sources, and protect citizens from online harms. The CMA also advocates for solutions that improve access to care so Canadians can rely on the healthcare providers they trust.
The survey has a margin of error of plus or minus 1.38%, 19 times out of 20. The findings raise urgent questions about how Canada’s healthcare system can adapt to ensure patients have access to reliable care and information as AI becomes increasingly prevalent in health-related contexts.
The Canadian Medical Association’s Health and Media Annual Tracking Survey was conducted online by Abacus Data from Nov. 3-13, 2025, with 5,000 Canadians participating.
About The Author
