A report described a family’s case alleging that ChatGPT advised an epilepsy patient to reduce medication before his death. It said GPT-4o repeatedly reassured him, but the reported official cause was sudden unexpected death in epilepsy; medical causation between the advice and death was not established. OpenAI expressed condolences and pointed to retirement of the consumer GPT-4o model and improvements in successors. Confident chatbot answers cannot evaluate an individual’s seizure risk or prescription. Decisions to reduce or stop medication should be made with the treating clinician.
Back to Tech
Medication advice from a chatbot faces scrutiny after a death
Repeated reassurance is not a substitute for clinical judgment.