Asger Risom
Google has disabled its AI overview after an investigation labelled the feature “dangerous”.
Google has disabled its AI-generated overviews for a number of medical queries after an investigation found the feature was surfacing misleading — and in some cases dangerous — health information.
Earlier this month, The Guardian reported that Google’s AI overviews were providing incorrect medical advice in response to common health-related searches. In one example experts described as “really dangerous,” the AI advised people with pancreatic cancer to avoid high-fat foods — the opposite of standard medical guidance, and advice that clinicians warned could increase the risk of death.
In another case, the system returned false information about liver function tests, potentially leading people with serious liver disease to believe their results were normal.
Medical queries now return no AI summary
As of this Sunday, AI overviews no longer appear for questions such as “what is the normal range for liver blood tests?” Instead, Google now falls back to traditional search results without an AI-generated summary.
Google declined to comment directly to The Guardian on the removals. In a statement to The Verge, spokesperson Davis Thompson said the company “invests significantly in the quality of AI Overviews, particularly for topics like health,” adding that internal clinicians had reviewed the examples raised.
According to Google, many of the flagged responses were “supported by high quality websites,” though the company acknowledged that AI overviews can miss context and said it takes action under its policies when necessary.
A growing pattern of AI search failures
The rollback is the latest in a string of controversies surrounding Google’s AI search features. Since launch, AI overviews have been shown suggesting people put glue on pizza, eat rocks, and follow other patently unsafe advice. The feature has also been cited in multiple lawsuits.
Health searches are especially sensitive because users often treat search results as authoritative. Even small inaccuracies can have outsized consequences when they influence medical decisions.
For now, Google appears to be taking a cautious step back — at least for some health-related queries — as scrutiny around AI-generated answers continues to grow.
Sources: The Guardian, The Verge, Google
