Sciencepulse

A Made-Up Disease Fooled the Chatbots

Health AI·November 8, 2026

More people are asking AI chatbots to explain their medical test results and symptoms. A researcher wanted to see what happens when those tools are fed a falsehood, so she invented one.

Almira Osmanovic Thunström created a fake disease she called "bixonimania." The chatbots took it in easily and repeated it to users, presenting the invented condition as a real medical matter. Osmanovic Thunström explained the experiment to Rachel Feltman on Science Quickly, the Scientific American podcast.

The concern is not only that a chatbot can be wrong. It is that a chatbot can state a falsehood in the same calm, authoritative tone it uses for accurate information. A reader who has no medical training has little to go on when deciding whether to trust the answer.

That risk matters because the habit is spreading. A recent survey cited in the episode found that a third of Americans say they have asked AI to decode their medical results. Nature covered the experiment in April 2026, and its reporting drew the same lesson: AI systems can take in and repeat claims that have no basis in medicine.

For patients, the practical takeaway is simple. An AI explanation can help you prepare questions for a doctor, but it should not replace a conversation with a clinician about what a result means and what to do next.