OpenAI Launched a Health Chatbot Feature One Day After a Lawsuit Alleged Its Advice Nearly Killed a User

OpenAI Expanded Health in ChatGPT Nationwide One Day After a Lawsuit Alleged It Nearly Killed a User

OpenAI opened Health in ChatGPT to every U.S. user over 18 on July 23 1. The day before, a former Florida pastor named Scott Winters sued the company and CEO Sam Altman in San Francisco Superior Court, claiming ChatGPT-4o had talked him out of seeking emergency care until a pulmonary embolism nearly killed him 2. The complaint asks a California court to halt Health in ChatGPT until independent auditors verify it is safe 2.

OpenAI expanded the product to the full country the following morning.

Health in ChatGPT lets users connect data from Apple Health, MyFitnessPal, One Medical, Function Health, and hospital systems running Epic or Oracle Health, then tag a message with @Health to pull that information into any conversation 1. Users can ask the chatbot to compare lab results against prior tests, summarize what changed since a doctor visit, or explore how sleep and activity track together 2. The rollout covered free, Go, Plus, and Pro tiers on web and iOS 2. OpenAI built the feature with input from hundreds of physicians who wrote realistic health scenarios and scoring rubrics, and says connected data is encrypted, never used to train its models or target ads, and deleted within 30 days of disconnection 2.

The demand is already there. OpenAI says users ask 300 million health-related questions every week, and that 70% of those queries took place outside the dedicated health hub during testing 1. Expanding @Health to all conversations formalizes behavior the company was already seeing.

According to the complaint, ChatGPT-4o told Winters his dizziness and blood pressure swings were "not something dangerous" and dismissed them as part of a minor, ongoing pattern 2. The chatbot advised him to stay in his recliner and said he would need eight to ten more episodes before the symptoms warranted serious concern 2. Hours before he collapsed, when Winters flagged groin pain, the chatbot responded with religious language: "God did not design your body to endlessly fail" 2. He delayed treatment for roughly six weeks and suffered a massive pulmonary embolism in July 2025 2. The collapse cost him his job and his ministry, and eventually his home 2.

The complaint includes claims of strict liability, negligence, invasion of privacy, practicing medicine without a license, and violations of California's unfair competition law 2. Beyond monetary damages, Winters' legal team asks the court to mandate stronger health guardrails, pull GPT-4o, delete training data tied to the conduct, and halt Health in ChatGPT pending independent safety audits 2. He is represented by the Social Media Victims Law Center, Tech Justice Law, and Temple University's Institute for Law, Innovation and Technology 2.

The case is among the first to target ChatGPT specifically over medical advice 2. Meetali Jain, director of Tech Justice Law, said the chatbot inserts itself as a wedge between a user and their real-life support network 2. OpenAI spokesperson Drew Pusateri responded that ChatGPT is not a doctor and should never substitute for medical care, diagnosis, or treatment, and said reducing Winters' medical decisions to the chatbot alone oversimplifies the situation 2.

OpenAI's own terms of service state that its services are not intended for diagnosis or treatment of any health condition 1. Its launch announcement acknowledges that ChatGPT can still get medical questions wrong 2. The product now includes physician input, encrypted data handling, and more capable models than the one Winters interacted with. None of those additions change the fine print.

On July 23, OpenAI gave every adult in the country access to a health product whose own terms advise against relying on it for health. A court has been asked to stop it. OpenAI broadened it instead.