Artificial Intelligence

Study: ChatGPT Health Misses Half of Emergency Cases

Is ChatGPT Health a dangerous tool missing critical safety features or the future of accessible healthcare?

Published MAR 4

Study: ChatGPT Health Misses Half of Emergency Cases

Story

Above: The Torch Health and OpenAI logos. OpenAI had acquired Torch Health to develop ChatGPT Health. Image credit: Samuel Boivin/NurPhoto/Getty Images

The Facts

In a study published in Nature Medicine last week, an evaluation by researchers at the Icahn School of Medicine of ChatGPT Health — an AI tool launched by OpenAI in January — found that it under-triaged more than half of emergency medical cases, failing to direct users appropriately to emergency care in serious situations.
FOX News
Medicalxpress
Nature
According to OpenAI, approximately 40 million people use ChatGPT Health daily to seek health information and guidance, including advice about whether to seek urgent or emergency care. The tool allows users to connect their medical records to receive health advice and responses.
guardian
Medicalxpress
Openai
The research team created 60 structured clinical scenarios spanning 21 medical specialties, ranging from minor conditions to true medical emergencies. Three independent physicians determined the appropriate level of urgency for each case using guidelines from 56 medical societies, conducting 960 interactions with ChatGPT Health in total.
NDTV
Digital Health
Nature
While ChatGPT Health performed well in clear-cut emergencies such as stroke or severe allergic reactions, it struggled in more nuanced situations. In one asthma scenario, the system identified early warning signs of respiratory failure in its explanation, but still advised waiting rather than seeking emergency treatment.
FOX News
Digital Watch Observatory
The study identified serious concerns with ChatGPT Health's suicide-crisis safeguards, finding that alerts appeared inconsistently. The system's alerts were inverted relative to clinical risk, appearing more reliably for lower-risk scenarios than in cases when someone shared how they intended to hurt themselves.
NBC
Anadolu Agency
In response, OpenAI stated that the study did not reflect how people typically use ChatGPT Health or how the product is designed to function in real-world health scenarios. The company noted that the model undergoes continuous updates and refinements, adding ChatGPT Health is currently available to a limited number of users.
NBC
Digital Health

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Techno-skeptic narrative

ChatGPT Health fails to recognize emergencies when lives hang in the balance, missing half of urgent cases and dangerously advising patients to wait during respiratory failure and diabetic crises. The tool creates a false sense of security that could cost lives, particularly when it inconsistently alerts high-risk suicide cases while firing warnings for lower-risk scenarios. These consumer-facing AI systems need rigorous monitoring and safeguards before millions begin relying on them for critical health decisions.

You Can Know Things

New York Post

Techno-optimist narrative

AI is the future of healthcare as millions already use ChatGPT for health queries, gaining clarity before consulting doctors and accessing research-style medical breakdowns. The integration of personal health data with AI creates a powerful synthesis that transforms these tools into true partners rather than simple question-answering systems. This personal data ecosystem marks a major step toward making healthcare more accessible and informed.