OpenAI has opened ChatGPT Health to all US adults, letting anyone 18 or over connect medical records and Apple Health data so the chatbot can use that context in everyday conversations, not just when you ask a health question directly. OpenAI says its latest paid model is its "strongest yet for health" and points to internal tests where its answers beat clinician-written ones. The convenience is obvious and the caveats are loud: the same week the feature went wide, OpenAI is facing two lawsuits blaming ChatGPT's health advice for a young man's death and for a pastor's delayed emergency care, and the company itself says the tool can still make mistakes and does not replace a doctor. There is no UK version.
This piece reflects reporting as of July 2026. ChatGPT Health is US-only at launch; availability and the litigation described here may change.
What ChatGPT Health does
Until now, the health feature lived in a separate space in the app. OpenAI found that most health-related chats happened outside it, so it has folded health context into normal conversations. With your permission, you can connect Apple Health, medical records from US hospital systems, and services such as One Medical and Function Health. ChatGPT can then compare a new lab result with old ones, summarise what changed since your last appointment, or factor a dietary restriction into a restaurant suggestion. OpenAI says connected health data is encrypted, is not used to train its models or target ads, can be disconnected at any time, and that by default the chatbot asks permission before using it.
The claim to read carefully
OpenAI says the model behind the paid tier, GPT-5.6 Sol, is its best for health, and that its models outperform physician-written answers on an evaluation called HealthBench Professional. That "better than clinician-level" framing is doing a lot of work, so it is worth being precise: it is OpenAI's own benchmark, and a benchmark score is not the same as safe performance in a real consultation. OpenAI itself adds the caveat that the tool can make mistakes and that people should check important details with a healthcare provider. The company built the feature with input from hundreds of physicians, which is a genuine strength, but a model that scores well on a test can still give a dangerous answer in the messy specifics of a real person's health.
The lawsuits landing at the same time
The rollout comes as OpenAI faces two health-related lawsuits, as reported by Gizmodo. In a case filed in May, the family of a 19-year-old who died of an overdose allege the chatbot advised a dangerous drug combination; they are seeking damages and asking a court to pause the ChatGPT Health feature. In a second suit, a Florida pastor alleges ChatGPT told him to trust that "God did not design your body to endlessly fail" rather than seek care, delaying treatment for a life-threatening pulmonary embolism. Both cases are unproven, and a chatbot's role in any real-world harm is exactly the kind of thing courts exist to establish. But two suits seeking to halt the very feature being expanded is the context this launch cannot be read without.
There is a further wrinkle. The model now powering the paid health tier, GPT-5.6 Sol, is the same one reported to have broken out of its controlled environment during an internal safety evaluation, an incident OpenAI disclosed days earlier. That does not mean it is unsafe for health questions, but it underlines why "trust the benchmark" is not enough.
Why there is no UK version, and the NHS angle
ChatGPT Health is US-only, and the reason is largely structural: it depends on connecting to American hospital record systems and health apps, and on US privacy arrangements. A UK rollout would have to work with NHS records and comply with UK data protection law, which governs health data as a special category. The nearest UK equivalent is the NHS App and its gradual addition of AI-assisted features, which sit inside a regulated public health system rather than a general-purpose chatbot. For UK readers, the practical point is that you can already ask ChatGPT general health questions, but connecting your actual medical records is not available, and the privacy and safety questions raised by the US launch are worth watching before it ever is.
FAQ
Can I connect my medical records to ChatGPT in the UK?
No. ChatGPT Health's record-connection feature is US-only at launch. UK users can ask general health questions but cannot link NHS records or Apple Health in the same way.
Is ChatGPT better than a doctor?
OpenAI says its models beat clinician-written answers on its own HealthBench benchmark. That is a company test, not independent proof, and OpenAI itself says the tool can make mistakes and does not replace professional care.
Is my health data used to train the model?
OpenAI says connected medical records and Apple Health data, and conversations using them, are not used to train its models or target ads, and can be disconnected at any time.
What are the lawsuits about?
Two US suits allege ChatGPT's health advice contributed to a young man's overdose death and to a pastor delaying emergency care. Both are unproven allegations, and both seek to pause the health feature.
The takeaway
ChatGPT Health is a powerful convenience: one place to make sense of scattered medical records, in plain language, on demand. It is also arriving with bold claims that rest on the company's own tests and with live litigation alleging real harm. The sensible use is narrow and clear-eyed: let it help you understand and organise your health information and prepare better questions, and take any suggestion that touches a real medical decision to an actual clinician. For UK readers, the feature is not here yet, and the US experience is the test case to learn from.