Overview
- A Guardian investigation documented inaccurate advice in AI Overviews, including a recommendation for pancreatic cancer patients to avoid high‑fat foods, incorrect information about women’s cancer tests, and misleading liver blood test ranges.
- Mental health organizations said some summaries on psychosis and eating disorders were harmful and could discourage people from seeking help.
- Google maintains the vast majority of Overviews are accurate, says they link to reputable sources and are reviewed by internal clinicians, and provided case‑by‑case clarifications in response to the reported examples.
- ZDNET reviewed Google’s replies and ran limited tests, observing more qualified language in some answers and notable variability based on how queries were phrased.
- Surveys and studies from Annenberg and MIT show many people seek health information online and often trust AI outputs, heightening the risk when summaries omit context or err.