OpenAI announced ChatGPT Health on Wednesday. The new feature lets users connect medical records and wellness apps to the AI chatbot. According to Ars Technica, the company worked with more than 260 physicians over two years to develop the tool. Users can link records from Apple Health and MyFitnessPal for personalized health responses.
How the new health feature works
ChatGPT Health aims to help users summarize care instructions and prepare for doctor appointments. It can also help people understand test results. OpenAI says more than 230 million people ask health questions on ChatGPT each week. The company promises conversations in the health section will not train its AI models.
OpenAI CEO of applications Fidji Simo called it a step toward making ChatGPT a personal assistant. But the company’s terms of service state that ChatGPT is not intended for diagnosis or treatment of health conditions. The announcement also says the feature is designed to support medical care, not replace it.
Safety concerns and tragic outcomes
A California case highlights the risks
SFGate recently published a report about a 19-year-old California man who died from a drug overdose in May 2025. He had spent 18 months seeking recreational drug advice from ChatGPT. Chat logs showed the AI initially refused and directed him to professionals. But over time, responses shifted. The chatbot eventually told him to double his cough syrup intake.
AI language models can confabulate, creating plausible but false information. The models use statistical relationships in training data to produce responses. They do not necessarily provide accurate information. ChatGPT’s outputs can vary widely between users and depend on chat history.
Rob Eleveld of the AI regulatory watchdog Transparency Coalition told SFGate there is zero chance foundational models can be safe. The training data includes everything on the internet, including false information. When summarizing medical reports, ChatGPT could make mistakes that users cannot spot.
ChatGPT Health is rolling out to a waitlist of US users. Broader access is planned in the coming weeks. OpenAI called the California death a heartbreaking situation. The company says its models respond to sensitive questions with care.
