Google Pulls AI Health Overviews for Select Medical Queries

Google has reportedly scaled back the use of its AI-generated health summaries on Search after concerns were raised over the accuracy of the information being displayed to users.

According to a report by The Guardian, the company has removed its AI Overviews for certain medical-related queries, including searches such as “what is the normal range for liver blood tests” and “what is the normal range for liver function tests.” The action follows an investigation by the publication which found that the AI-generated summaries were producing information that was either misleading or lacking crucial medical context.

Responding to the development, a Google spokesperson told the publication that the company does not comment on individual removals within Search. However, the spokesperson added that when AI Overviews fail to provide adequate context, Google works on broader improvements and takes corrective action in line with its internal policies.

This is not the first time concerns have been raised over AI-generated medical summaries. A previous investigation by The Guardian highlighted multiple instances in which AI Overviews appearing at the top of search results delivered inaccurate health-related information, potentially putting users at risk. Experts noted that the AI’s responses often varied widely from what is clinically considered “normal.”

In one example cited by the report, a search for liver blood test ranges produced long lists of figures without sufficient explanation, while ignoring key variables such as age, sex, ethnicity, or nationality — all of which are essential in interpreting medical test results accurately.

Health sector representatives have welcomed Google’s decision to remove the problematic summaries. Vanessa Hebditch, Director of Communications and Policy at the British Liver Trust, described the move as “excellent news,” saying she was pleased that Google had withdrawn AI Overviews in these specific cases.

At the same time, experts cautioned that the issue has not been fully resolved. Hebditch warned that if similar questions are phrased differently, misleading AI-generated summaries could still appear, raising ongoing concerns about the reliability of automated health information.

Echoing those views, Sue Farrington, Chair of the Patient Information Forum, welcomed the removal but stressed that it is only an initial step. She told The Guardian that maintaining public trust in health-related search results would require far more rigorous oversight, noting that several examples of inaccurate AI Overviews remain accessible.

The episode has renewed scrutiny on the growing role of artificial intelligence in delivering health information online and the responsibility of tech platforms to ensure accuracy when users turn to search engines for medical guidance.