Google removes AI Overviews for specific medical queries

Google appears to have removed AI Overviews—the company’s AI-generated search summaries—for some health-related queries. The change follows an investigation by UK newspaper The Guardian, which reported that the search feature was offering misleading information in response to certain medical questions. For instance, the publication noted that queries regarding normal ranges for liver blood tests returned summaries that omitted critical demographic factors like age, sex, ethnicity, or nationality, potentially leading users to misinterpret their health status and believe their results were healthy when they were not.

While Google seems to have taken down the summaries for specific phrases like “what is the normal range for liver blood tests” and “what is the normal range for liver function tests,” The Guardian reported that variations on those queries could still lead to AI-generated summaries. For example, search terms such as “lft reference range” or “lft test reference range” could still trigger the automated overviews. Subsequent tests of these variations showed that while some queries no longer generated the summaries, Google still provided options to run the queries in AI Mode. A Google spokesperson declined to comment on individual removals within Search, but stated that the company works to make broad improvements. Additionally, the spokesperson noted that an internal team of clinicians reviewed the highlighted queries and found that in many instances, the information was not inaccurate and was also supported by high-quality websites.

The British Liver Trust, a health charity, welcomed the development. Vanessa Hebditch, the organization’s director of communications and policy, called the removal excellent news. However, she expressed concern over the piecemeal approach to the problem. “Our bigger concern with all this is that it is nit-picking a single search result and Google can just shut off the AI Overviews for that but it’s not tackling the bigger issue of AI Overviews for health,” Hebditch said. She warned that addressing individual search results does not solve the broader issue of AI reliability in health contexts, emphasizing that simply disabling summaries for specific queries does not address the broader issue of AI Overviews for health.

Why it matters

This incident underscores the ongoing challenge for tech giants in balancing the rapid deployment of generative AI with the high-stakes requirement for accuracy in medical information.