When you do an Internet search for health information in Google, you will see summary boxes popping at the very top of the search results page, generated by artificial intelligence technology, but what worries us most is that it references YouTube even before any medical-related site, according to recent research in health information literacy.
One study investigated more than 50,000 web searches for health-related topics. It found that “YouTube dominated as a first source for Google’s AI-generated answers, representing 4.43% of all citations.”
This may not be a very strong number but “exceeded all hospital networks, government health sites, medical associations, and educational institutions combined.”
The investigation, which used the search engine optimization platform, also known as SE Ranking, searched for German language health queries to determine which sources the Google tool trusted the most, and what the findings reveal raises serious concern for health experts with regards to how the Internet will influence health news in the future.
YouTube Outranks Medical Authorities
The numbers tell a striking story. Out of 465,823 total citations in AI overviews, YouTube accounted for 20,621 citations. Coming second was NDR.de, a German public broadcaster, which accounted for only 14,158 citations. A medical reference website Msdmanuals.com accounted for a paltry 9,711 citations as a third source.
This is important to remember because, despite how it may appear, YouTube is not a medical publisher. Anyone, including certified physicians or hospital channels, wellness evangelists, life coaches, or total non-experts with zero medical background or experience, can publish medical content on this platform.
“This matters because YouTube is not a medical publisher,” the researchers underscored. “It is a general-purpose video platform.”
This comes after an investigation by the Guardian revealed that persons were being endangered by false and misleading health information provided in response to their queries in Google’s artificial intelligence feature. There was an instance in which it is believed that persons with presently serious liver disease may have thought they were perfectly healthy after being wrongly told this by Google in response to their queries on liver function tests.
Google’s Defense Falls Short
Google responded by saying that Google AI Overviews is intended to feature quality content from credible sources irrespective of their formats. Google further mentioned that there are many credible sources on healthcare issues and medical content created by certified medical professionals on YouTube.
However, once again, this fact was challenged by the researchers on the basis of a ‘cold hard reality check’, whereby they mentioned that only “the top 25 videos constitute less than one percent of all YouTube links cited by AI Overviews on health-related topics.”
“So at first glance it looks pretty reassuring,” the researchers noted. “But it’s important to remember that these 25 videos are just a tiny slice. With the rest of the videos, the situation could be very different.”
A Structural Problem, Not Isolated Incidents
Hannah van Kolfschooten, a researcher specializing in AI, health, and law at the University of Basel, said the study provides crucial evidence that the risks aren’t just occasional mishaps.
“This study provides empirical evidence that the risks posed by AI Overviews for health are structural, not anecdotal,” she explained. “It becomes difficult for Google to argue that misleading or harmful health outputs are rare cases.”
Van Kolfschooten emphasized that the heavy reliance on YouTube over public health authorities suggests that visibility and popularity drive AI Overviews, rather than medical reliability.
The researchers conducted their study in Germany specifically because its healthcare system operates under strict regulations from German and EU directives. Their reasoning: if AI systems rely heavily on non-authoritative sources even in such a regulated environment, the problem likely extends far beyond a single country.
AI Search Risks Global Health Literacy by Favoring YouTube Trends Over Verified Data
AI Overviews are utilized on more than 82% of health-related search inquiries, as per the study. This means that every month, about 2 billion individuals have the possibility of coming across AI-generated health content that gives precedence to popular video content over medical content.
The study’s researchers had even qualified their study by stating that it was just a snapshot at a particular date in December 2025, focusing only on queries in the German language and that results may vary from region to region depending upon the wording of the queries.
Such study results evoke serious questions regarding the impact of artificial intelligence tools on public information in health matters.
The issue, of course, is no longer simply is it Smart Search is efficacious, but rather is it efficacious in guiding people towards information that is medically sound when they need it most?
