Recently, Google announced the deactivation of certain AI overview features related to health, following an investigation by The Guardian. The investigation found that Google's AI system provided misleading information when handling certain health issues. For example, when users searched for "normal liver blood test ranges," the values displayed by Google did not take into account individual differences such as nationality, gender, race, or age, which could lead users to incorrectly believe their test results were normal.
In this context, The Guardian reported that Google has removed AI overviews about "normal liver blood test ranges" and "normal liver function test ranges." However, for queries like "LFT reference ranges" or "LFT test reference ranges," Google still generates AI summaries. This has sparked widespread discussions about the accuracy of Google's AI health features.
A Google spokesperson told The Guardian that the company does not comment on the removal of individual search results but has committed to broader improvements. They emphasized that Google's internal clinical team has reviewed the relevant queries and stated that in many cases, the information provided was not inaccurate and was supported by high-quality websites. Nevertheless, users and experts remain concerned about the performance of AI in the health field.
Vanessa Hapgood, director of communications and policy at the UK Liver Trust, said in an interview that Google's removal of AI overviews is "good news." She further stated, "Our main concern is that Google has fine-tuned a particular search result, but the larger issue remains unsolved." This indicates that while Google has taken measures in some areas, its AI system still needs to be more cautious and accurate in the healthcare field.
Looking back at 2025, Google introduced a series of new features, including enhanced overviews and specialized health AI models, aimed at improving Google Search's application in the healthcare field. However, with this incident, Google needs to re-examine how to ensure the accuracy and reliability of health information provided, to avoid misleading users.
