Google has removed several AI Overviews that appeared for health-related searches after an investigation found that some of the summaries gave misleading or dangerous guidance. The Guardian reported tests showing summaries that did not account for age, sex, ethnicity, or individual medical context. Experts warned that some answers could harm patients who act on the advice without consulting clinicians.
Search summaries appear at the top of results and are designed to give quick answers. The feature uses Google’s Gemini models to synthesize information from multiple sources. The prominence of AI Overviews means that users may read the summary and not click through to the referenced sites. That pattern raised alarm among charities and medical experts because errors at that top position can carry outsized risk.
What was Removed
Reportedly removed results included an AI Overview for the query about normal ranges for liver blood tests. The summary presented ranges without noting differences by age, sex, or other factors. Another flagged result gave a broad recommendation about fat intake for people with pancreatic cancer. Charity representatives said that advice could harm patients who rely on it to make treatment-related choices. Health groups pointed to the need for clear context on lab values and for safe guidance on diet during cancer care.
The Guardian worked with charities and clinicians to test a range of medical queries and shared findings with Google. Other outlets then reported additional examples of AI Overviews that mischaracterised screening tests, and that may have given incorrect steps for care. The pattern of errors showed that some overviews trimmed important details in ways that experts found worrying.
Google Response Review
Google acknowledged the reporting and said it does not comment on individual removals. The company said it makes broad improvements when overviews miss context and that it has an internal clinical review team. Google said many of the flagged examples link to respected sources and that clinicians reviewed the cases shared by reporters. The company also stressed that it continues to refine models and to invest in quality controls for health queries.
Health charities and clinicians pushed back on Google statements. Pancreatic Cancer UK called one AI Overview on pancreatic cancer advice completely incorrect and warned that following the summary could jeopardise a person’s ability to receive treatment. Several medical groups said that prominent AI-driven summaries must meet a high bar for clinical accuracy and must clearly signal uncertainty.
