Google has quietly removed its AI Overviews feature from some medical search results after reports showed that the tool was giving incorrect and potentially harmful health advice.
The decision follows an investigation by The Guardian that raised serious concerns about how Google’s AI was handling sensitive medical questions.
In one example highlighted by experts, Google’s AI wrongly told people with pancreatic cancer to avoid high-fat foods. Doctors said this advice was extremely dangerous because patients with this condition are often encouraged to consume higher-fat diets to maintain weight and strength. Giving the opposite advice, experts warned, could increase the risk of death.
Another troubling case involved liver blood tests. The AI provided false information about normal test ranges, which could lead people with serious liver disease to believe their results were healthy when they were not. Medical professionals described these answers as alarming, since misunderstandings like this could delay diagnosis or treatment.
As of January 12, users searching for questions such as “what is the normal range for liver blood tests?” no longer see AI Overviews at all. Instead, Google appears to have disabled the feature entirely for certain medical queries. Google declined to comment directly on why specific searches were changed.
However, a Google spokesperson told The Verge that the company invests heavily in improving the quality of AI Overviews, especially for health-related topics. According to Google, most of the information provided by the feature is accurate and backed by high-quality websites. The company said its internal team of clinicians reviewed the reported examples and found that, in many cases, the information was not wrong but lacked important context. Google added that when problems are identified, it works to improve the system and takes action based on its internal policies.
This latest issue adds to a growing list of controversies surrounding Google’s AI Overviews. In the past, the feature has been criticized for suggesting unsafe or bizarre actions, such as telling users to put glue on pizza or eat rocks. It has also been linked to multiple lawsuits, raising broader questions about the risks of using generative AI to answer questions where accuracy is critical.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
The removal of AI Overviews from some medical searches suggests that even Google recognizes the limits of current AI technology when it comes to health information. For now, it appears the company is taking a step back in areas where mistakes could have serious real-world consequences.
