AI Chatbots Steer Health Queries to Stem-Cell Clinics
A peer-reviewed study finds the commercial sources chatbots cite can route patients toward unproven treatments, a research-stage finding.
Updated on Nov. 6, 2026 in Artificial Intelligence
Live Poll
Do you trust AI chatbots to give you reliable health information?
Six generative AI chatbots, including ChatGPT and Meta's Llama, cited for-profit providers of unproven stem-cell interventions when answering health questions about Alzheimer's disease, autism and cerebral palsy, according to a study from Simon Fraser University published in BMJ Connections Digital Health & AI. This is a research-stage finding based on analysis of cited references, not of the answers themselves.
Why it matters
The study shows that AI chatbots can steer health-seeking users toward commercial clinics selling interventions without proven benefit, a risk that grows as people use these tools for medical questions. It arrives alongside a second SFU peer-reviewed study that found AI-generated fake testimonials on medical clinic websites, suggesting the commercial medical marketplace is adapting to generative AI.
The study measured 5,000-plus references across 540 queries, with for-profit providers accounting for 23.4 per cent of citations and stem-cell providers nearly 83 per cent of those commercial citations. Because the researchers analyzed cited resources rather than the answer text, the figures describe what the chatbots pointed to, not what they told users.
The players
Simon Fraser University
A Canadian research university whose researchers conducted the study and the related study on AI-generated fake testimonials.
ChatGPT
OpenAI's generative AI chatbot, one of the six models tested for citing for-profit stem-cell clinics.
Meta's Llama
Meta's family of open-weight large language models, one of the six chatbots tested in the study.
The details
Researchers at Simon Fraser University asked six generative AI chatbots common questions about conditions linked to stem-cell treatment, including Alzheimer's disease, autism and cerebral palsy. The queries were run in five locations across North America and Europe, in English and French. The team then counted and classified the sources each chatbot cited, rather than evaluating the wording of its answers, so the finding concerns which commercial providers appeared as references.
Timeline
Study published in BMJ Connections Digital Health & AI on November 6, 2026.
The Tech Race
The same university's earlier peer-reviewed work on AI-generated fake testimonials on clinic websites points to a pattern in which commercial medical marketers use generative AI. This study moves the question from marketing copy to the answers and citations that general-purpose chatbots produce for patients.
Anyone asking a chatbot about Alzheimer's, autism or cerebral palsy may be shown commercial clinics that sell unproven stem-cell treatments as references. Readers should treat cited clinics as marketing sources and check claims against medical guidance before paying for any intervention.
The takeaway
As chatbots become a default first stop for health questions, the sources they cite will matter as much as their answers. Watch for whether the researchers or other groups extend this work to analyze the answer text itself, which the current study did not do.
Further reading
For more on how these tools are reshaping medical information, explore our Artificial Intelligence coverage.
Source note: This article includes information reported by The Vancouver Sun.
Live Poll
Do you trust AI chatbots to give you reliable health information?
