OpenAI opens ChatGPT Health to all US users as scrutiny mounts

OpenAI has removed the last gate on its health product. ChatGPT Health, previously a feature tucked inside paid tiers, is now live for every US user aged 18 and over on Free, Go, Plus and Pro, rolling out on web and iOS during the week of 23 July 2026. That turns a niche hub into a mass-market entry point for medical questions, at zero marginal cost to the user.

The product, launched in January, lets people connect Apple Health, MyFitnessPal, Function and medical-record systems including Epic and Oracle Health. The bigger shift in this update is scope: health data now feeds every ChatGPT conversation, not just the dedicated hub. OpenAI's own testing found 70% of health questions were already happening outside the hub, meaning the company is now building the product around where usage actually occurs rather than where it was designed to occur.

Volume backs up the bet. Weekly health-related questions have grown from 230 million in January to 300 million, according to TechCrunch. OpenAI also says its newest model, GPT-5.6 Luna, outperforms GPT-5.5 on the HealthBench evaluation, giving it a fresh accuracy claim to lean on as it pushes deeper into a regulated category.

OpenAI says health data is not used to train its foundation models and that it works with physicians on response safety. Its terms state the service is "not intended for use in the diagnosis or treatment of any health condition" — a disclaimer that sits awkwardly against a free-tier rollout aimed at maximising exposure to medical queries.

The timing is pointed. The expansion lands as a Florida pastor sues OpenAI alleging ChatGPT gave dangerous medical guidance, a case that will test exactly the gap between OpenAI's safety claims and its liability exposure.

Opening Health to every free user is a distribution play: OpenAI is betting that habitual, everyday medical questions become a durable reason to open the app, ahead of Google and any consumer health assistant rivals might ship. The 300-million weekly figure shows the habit is already forming. But making Health ambient across all chats, not contained in a hub, also widens the surface area for exactly the kind of claim now in court. OpenAI is scaling reach and legal exposure at the same time, and the lawsuit's outcome will show whether "not intended for diagnosis" is a disclaimer that holds up once the product is this far into the main flow.