AI Chatbots Could Steer You Wrong on Health Issues, Fresh Research Shows — Hive
New Oxford University research indicates that AI chatbots frequently provide incorrect or inconsistent health advice, potentially leading to harm.Summary
A recent study from Oxford University, led by Dr. Rebecca Payne and published in Nature Medicine on February 9, 2026, found that popular AI chatbots often deliver wrong or mixed-up medical advice when users input symptoms. The research, which involved testing the models on real patients, revealed that the AI's guidance was inconsistent, sometimes failing to flag urgent care needs. Compared to a standard Google search, the chatbots offered no superior diagnostic accuracy or guidance on next steps. Experts involved in the trial, the largest of its kind on AI handling personal symptoms, concluded that these systems are not yet ready for use in patient care, noting that minor variations in how questions are phrased can drastically alter the advice given.
(Source：Hive)