A day after a Florida man sued OpenAI, alleging its chatbot gave him 'extremely dangerous medical recommendations' that nearly killed him, the company unveiled a feature that allows ChatGPT to access users' Apple Health data — including medical records, medications, and lab results.
The lawsuit, filed Wednesday in San Francisco Superior Court, claims ChatGPT repeatedly reassured the plaintiff that his symptoms — later diagnosed as a pulmonary embolism — were not serious and urged him to remain at home. He was subsequently hospitalised and now faces years of intensive physical and psychological recovery, according to the filing.
“OpenAI launches ChatGPT health records feature a day after lawsuit alleges chatbot advice caused near-fatal embolism.”
OpenAI’s new 'Health in ChatGPT' feature, announced Thursday, lets eligible users connect the chatbot to Apple’s Health app, giving it access to information such as sleep, activity, medications and allergies. The company says the integration provides 'additional context' for health-related conversations, allowing more personalised responses based on a user’s data.
But the rollout has renewed concerns about the safety of AI-driven health advice. These are separate from, but echo, worries highlighted by the BBC’s Tech Life programme about 'shadow banning' — a practice that may be limiting access to health education messages for women and young girls.
OpenAI insists ChatGPT 'is not a doctor and should never be used as a substitute for qualified medical professionals.' A spokesperson told The New York Times that the chatbot’s terms of service make clear it is not intended for medical diagnosis or treatment, and added that the company’s models have improved since the incident at the heart of the lawsuit.
The company said more than 70 percent of health-related conversations during earlier testing occurred outside a dedicated health space, prompting the decision to integrate health data into general chats. OpenAI also said it will not use connected health data to train its foundation models or to serve ads.
Despite these assurances, the timing of the launch — one day after a lawsuit alleging near-fatal consequences — has raised questions about whether the safeguards are sufficient. The plaintiff claims the chatbot’s advice discouraged him from seeking medical care as his condition worsened, leading to a near-fatal embolism.
OpenAI says it worked with hundreds of physicians to test the feature for performance and safety, but acknowledges that 'ChatGPT is not a doctor and should never be used as a substitute for qualified medical professionals.' The question remains whether users will heed that warning.