Google AI Overviews cite YouTube more than any medical site for health queries, study suggests
A German study of over 50,000 health queries shows Google’s AI Overviews feature cites YouTube more often than traditional medical websites. The findings highlight potential risks of relying on a platform where anyone can upload content, prompting calls for greater scrutiny of AI‑generated health i…
Google’s AI‑driven Overviews, the concise answer boxes that appear at the top of search results, are now under scrutiny after a German research team discovered that the feature references YouTube more frequently than any established medical website when answering health‑related questions. The analysis, which examined 50,807 health‑focused prompts captured from Berlin‑based searches, revealed that YouTube accounted for 4.43% of all citations – a higher share than any hospital network, government health portal, or academic institution.
How the study was conducted
Researchers from SE Ranking, a platform specializing in search‑engine optimisation analytics, collected a snapshot of AI Overview responses in December 2025. They focused on German‑language queries because Germany’s healthcare system is tightly regulated by both national and EU directives, offering a stringent backdrop for evaluating source credibility. The team parsed the underlying URLs that the AI cited for each answer, tallying the frequency of each domain across the 50,807 health‑related prompts.
Overall, AI Overviews appeared in more than 82% of the health searches examined, indicating that the feature is a dominant entry point for German users seeking medical information online. The most cited domain was YouTube, with 20,621 citations out of a total of 465,823 source references. The next most common source was the public broadcaster NDR.de (14,158 citations), followed by the medical reference site Msdmanuals.com (9,711 citations), the consumer health portal Netdoktor.de (7,519 citations), and the doctor‑career platform Praktischarzt.de (7,145 citations).
Why YouTube’s prominence is concerning
YouTube is a general‑purpose video platform that allows anyone to upload content, ranging from board‑certified physicians and hospital channels to wellness influencers and unqualified creators. While some of the cited videos are produced by reputable medical institutions, the sheer volume of links means that a substantial portion could originate from sources lacking medical expertise. Critics argue that the AI Overview algorithm appears to prioritize visibility and engagement metrics over the rigor of the source, potentially exposing users to misinformation.
Google has responded by emphasizing that the AI Overviews are designed to surface “high‑quality content from reputable sources, regardless of format.” The company points out that the 25 most‑cited YouTube videos in the study were 96% from medical channels, and 21 of those 25 explicitly noted that the content was created by licensed professionals. However, SE Ranking researchers caution that these 25 videos represent less than 1% of all YouTube links cited, leaving the majority of video references unexamined.
Broader implications for AI‑generated health advice
The findings echo earlier concerns raised by a Guardian investigation, which documented cases where Google’s AI Overviews delivered inaccurate or dangerous health information—such as erroneous liver function test results that could mislead patients with serious liver disease. Following that investigation, Google removed AI Overviews for some medical queries but retained the feature for others, sparking debate over the consistency of its safety measures.
Hannah van Kolfschooten, a researcher at the University of Basel who specializes in AI, health, and law, noted that the German study provides “empirical evidence that the risks posed by AI Overviews for health are structural, not anecdotal.” She argued that the reliance on YouTube suggests the system rewards popularity and accessibility over medical authority, making it harder for Google to claim that harmful outputs are isolated incidents.
Limitations and next steps
The study’s authors acknowledge several constraints. It captures a single moment in time, focuses exclusively on German‑language queries, and does not account for regional variations in source availability or user phrasing. Consequently, the results may differ in other countries or languages. Nonetheless, the researchers argue that if a heavily regulated market like Germany shows a bias toward non‑medical platforms, similar patterns could exist elsewhere.
Google’s spokesperson countered that the report’s own data shows the “most cited domains are reputable websites,” and that the AI Overviews cite expert YouTube content from hospitals and clinics. The company maintains that its AI models continuously learn from user feedback and that it will refine citation algorithms to prioritize authoritative medical sources.
For users, the takeaway is clear: while AI Overviews offer quick answers, they should not replace professional medical advice. Health seekers are encouraged to verify information against trusted sources such as the Centers for Disease Control and Prevention (CDC), the Mayo Clinic, or directly consult healthcare providers.
What may happen next
Regulators in the European Union are increasingly focusing on AI transparency and safety, particularly for high‑risk applications like health advice. The European Commission’s AI Act, slated for implementation in the coming years, could impose stricter obligations on platforms that generate medical content. In response, Google may need to adjust its citation weighting, enhance source verification, or provide clearer disclosures about the provenance of AI‑generated answers.
Meanwhile, independent researchers and consumer‑advocacy groups are likely to continue monitoring AI Overviews, expanding analyses to other languages and regions. Their work could pressure tech giants to adopt more rigorous standards for health‑related AI outputs, ensuring that the convenience of instant answers does not come at the expense of public safety.
Why it matters
The study shows that AI‑generated health answers may rely on less‑authoritative sources, raising the risk of misinformation for billions of search users.
Key points
- YouTube was the most cited source in Google AI Overviews for health queries in a German study
- AI Overviews appeared in over 82% of health‑related searches examined
- The top 25 YouTube videos were largely from medical channels, but they represent <1% of all video citations
- Google argues its AI cites reputable content, while critics say popularity outweighs reliability
- EU AI regulations could force Google to tighten source verification for medical answers
Frequently asked questions
What are Google AI Overviews?
AI Overviews are concise answer boxes generated by Google’s generative AI that appear at the top of search results, summarising information from multiple sources.
Why does the study focus on Germany?
Germany’s healthcare system is tightly regulated by national and EU rules, providing a stringent environment to test whether AI Overviews rely on authoritative medical sources.
Did Google remove AI Overviews after the Guardian report?
Google removed the feature for some medical queries but kept it for others, prompting ongoing debate about its safety policies.
Are the YouTube videos cited by AI Overviews reliable?
The study found that the 25 most‑cited videos were mostly from medical channels, but they make up less than 1% of all YouTube links used, leaving many citations unverified.
What could change the way Google cites sources for health answers?
Potential EU AI regulations, increased public scrutiny, and pressure from researchers may lead Google to adjust its citation algorithms to prioritise verified medical institutions.
