Study Finds ChatGPT Health Under-Triages Over Half of Emergency Medical Cases

A study published in Nature Medicine evaluated ChatGPT Health's triage recommendations using 60 clinician-authored medical scenarios across 21 clinical domains. The AI performed well in moderate cases and recognized many classic emergencies but under-triaged over half of gold-standard emergency cases, often advising less urgent care than needed. The research also noted inconsistent responses in mental health crisis scenarios. These findings raise safety concerns about relying on ChatGPT Health for emergency medical guidance.