Google AI Health Answers Lean Heavily on YouTube, Raising Accuracy Concerns: Study

The research, carried out by SEO analytics firm SE Ranking, examined more than 50,000 health-related searches in Germany. The findings showed that YouTube was the most frequently cited source in Google’s AI Overviews, appearing in 4.43% of citations. This was roughly three and a half times more than netdoktor.de, one of Germany’s largest consumer health websites, and more than double the citations attributed to MSD Manuals, a widely respected medical reference.

Researchers pointed out that YouTube is not a dedicated medical publisher. While the platform hosts content from licensed doctors, hospitals, and health organizations, it also includes videos from influencers and creators with no formal medical training. This mix, they argue, raises concerns when AI systems rely on such sources for health guidance.

According to the study, only about 34.5% of citations in AI Overviews came from what were classified as reliable medical sources. Academic journals and government health agencies together made up just around 1% of all citations, a figure that alarmed researchers.

Google has pushed back against the interpretation of the findings. In a statement to The Guardian, a company spokesperson said AI Overviews are designed to surface high-quality information from reputable sources across platforms. Google also noted that many verified health authorities and licensed professionals actively publish content on YouTube. The company added that the study had limitations, as it was based solely on German-language searches captured during a single period in December 2025.

The issue has drawn attention before. A previous investigation by The Guardian found that some AI-generated health responses contained misleading or incorrect information, potentially putting users at risk. Researchers involved in the SE Ranking study warned that if AI systems depend heavily on non-authoritative sources even in a well-regulated environment, the problem is likely broader than one country or language.

The findings have reignited calls for tighter oversight of health-related AI outputs. Google has already restricted AI Overviews for certain medical queries following earlier accuracy concerns.