Daijiworld Media Network - San Francisco

San Francisco, Feb 16: Google has removed some of its AI-generated search summaries after an investigation found misleading health advice that experts warned could put users at risk.

The controversy centred on Google’s AI Overviews feature, which had provided incorrect information about critical liver function tests. In some instances, users were reportedly given inaccurate “normal ranges,” raising concerns that patients with serious liver infections or diseases could mistakenly believe they were healthy.

Following the findings, Google removed AI Overviews for certain search terms, including queries such as “what is the normal range for liver function tests” and “what is the normal range for liver blood tests.”

A recent study by search engine optimisation platform SE Ranking further raised alarm, revealing that Google’s AI Overviews often relied on YouTube as a primary cited source for health-related queries rather than established medical websites.

The study, which analysed over 50,000 health searches in Germany, found that YouTube accounted for 4.43% of all AI citations — 3.5 times more than netdoktor.de, one of Germany’s largest consumer health portals, and more than twice the citations of the respected medical reference MSD Manuals. Only 34.45% of citations came from reliable medical sources, while government health institutions and academic journals together accounted for roughly 1%.

Researchers highlighted that YouTube is a general-purpose video platform, not a medical publisher, and hosts content from a wide spectrum of creators, including those without formal medical training.

Experts pointed to particularly dangerous examples. In one case, Google’s AI allegedly advised pancreatic cancer patients to avoid high-fat foods — advice medical professionals said was the opposite of recommended dietary guidance and could increase health risks. AI Overviews related to women’s cancer screening tests also reportedly contained incorrect information that might lead patients to dismiss genuine symptoms.

The rise of AI-driven health advice is not limited to search engines. According to OpenAI, nearly 40 million people globally use ChatGPT daily for healthcare-related queries. A 2026 survey by the Canadian Medical Association found that roughly half of Canadians surveyed consult Google AI summaries or ChatGPT for medical advice.

However, users who relied on AI-generated advice for self-diagnosis and treatment were reportedly five times more likely to experience adverse effects than those who did not.

Studies have also raised concerns about chatbot accuracy. A 2025 study by researchers at the University of Waterloo found that GPT-4 provided incorrect answers to open-ended health queries around two-thirds of the time. Another 2025 study by Harvard researchers suggested chatbots often failed to challenge flawed assumptions in user queries, instead offering compliant but misleading responses.

Health experts warn that while long wait times and limited access to doctors may drive patients toward AI tools for quick answers, overreliance on such systems without professional consultation could pose serious public health risks.

As AI tools become increasingly embedded in daily life, critics argue that ensuring accuracy, transparency and accountability in health-related information is more urgent than ever.