By Rebecca Ruiz
Publication Date: 2026-04-12 09:00:00
Every day millions of people turn to an artificial intelligence chatbot like Claude, Gemini and ChatGPT to ask a question about their physical health.
What you may not realize is that finding the right answer is harder than it seems, no matter how authoritative the chatbot answers.
Three recent studies suggest that large language models are not as reliable as users might hope.
Read this before using ChatGPT Health
A study testing chatbots’ ability to detect health misinformation failed more frequently in certain scenarios. Another study conducted by the same researchers found that ChatGPT Health, a dedicated health and wellness service launched in January, “underreported” just over half of the cases presented to it, including emergency situations that required immediate medical attention.
“I think consumers should have a high level of…