Source: The Guardian

The Guardian investigation has revealed that Google’s "AI Overviews" are providing false and misleading health information, potentially putting lives at risk. The feature, which uses generative AI to summarize search results, has been found to offer "really dangerous" advice, such as recommending that patients with pancreatic cancer avoid high-fat foods—the exact opposite of medical best practices. Other examples include providing bogus normal ranges for liver function tests and incorrectly listing pap tests as a screen for vaginal cancer. These errors could lead users to ignore critical symptoms or fail to follow up on life-saving medical appointments.

The report highlights that these AI-generated summaries often change with the same search query, leading to inconsistent and unreliable results. Mental health charities have also expressed deep concern, noting that advice for psychosis and eating disorders was frequently incorrect or harmful. While Google claims its summaries link to reputable sources, experts argue that the prominent placement of inaccurate AI text at the top of search results confuses consumers. The findings have intensified calls for stricter regulation of generative AI in high-stakes sectors like healthcare to prevent automated systems from disseminating life-threatening misinformation.