Photo: depositphotos
Google AI overview recommends dangerous diets for cancer patients, risking their health.
An investigation by The Guardian found that Google AI summaries can put people at risk. Early search results provided users with inaccurate or misleading advice on diagnosing and treating illnesses, prompting experts to warn that such guidance could lead to harmful decisions and delays in seeking medical care.
Google says its AI Overviews, powered by generative AI, are “helpful” and “reliable.” However, the investigation revealed multiple errors, particularly on health topics. In one “truly dangerous” case, Google advised people with pancreatic cancer to avoid high-fat foods—contradicting medical guidelines and potentially increasing the risk of death. Anna Jewell, director of support, research, and impact at Pancreatic Cancer UK, called the advice “completely wrong,” noting that following it could lead to calorie deficiency, weight loss, and the inability to tolerate chemotherapy or surgery.
Other problematic examples included misleading information on liver function tests, which could make seriously ill patients believe they were healthy and delay medical care. The AI also presented incorrect “normal ranges” for liver blood tests, without context or differentiation by age, sex, or ethnicity, making the results potentially dangerous. Pamela Healy, CEO of the British Liver Foundation, emphasized that “many liver disease patients show no symptoms until late stages, so testing is vital. What Google AI calls ‘normal’ can be dangerously misleading.”
The Guardian also found misleading summaries on mental health conditions. Steven Buckley from the charity Mind said some AI advice on psychosis and eating disorders was “incorrect, harmful, or could discourage people from seeking help.”
Athena Lamnisos, CEO of Eve Appeal, noted that AI summaries changed when the same search was repeated, producing different answers from various sources—meaning people could receive inconsistent guidance depending on timing.
Google responded that most AI summaries are accurate and useful, comparable to other search functions, and the company monitors cases where AI misrepresents information or ignores context. “We invest heavily in the quality of AI summaries, especially on topics like health, and the vast majority provide accurate information,” a Google spokesperson said.
