OpenAI wants to be in the room for your next diagnosis. On Thursday it began rolling out ChatGPT Health to every logged-in US adult.
The chatbot can now connect to your Apple Health data and medical records, the company said. It can then draw on them in any conversation. More than 300 million people already ask ChatGPT health questions each week.
Now it wants their lab results, medications and visit notes too.
The pitch came with a bold claim. OpenAI’s vice-president of health product, Ashley Alexander, made it plainly. Its models “are now capable of reasoning at levels that are better than clinician level,” she said, per The Verge.
Pressed on it, the company’s health lead Karan Singhal quickly softened the line. He would “temper” that, he said, though “individual studies” pointed that way. The headline claim did not survive the first follow-up question.
What it does #
The feature moves health out of a side tab and into the main chat. OpenAI tested a dedicated health space earlier this year, but more than 70% of health chats happened outside it, so it merged the two.
Connect Apple Health, One Medical, Function Health or a supported US hospital system. ChatGPT can then compare a new lab result with old ones. It can summarise what changed since your last visit, or weigh a dietary restriction when suggesting a recipe. It asks permission before using the data.
Free users get GPT-5.5 Instant. Paying users get GPT-5.6 Sol, which OpenAI calls its “strongest model yet for health”.
OpenAI leaned hard on privacy. Connected records, Apple Health data and the chats that use them are not used to train its models or target ads, it said, whatever your other settings.
Everything is encrypted, you can revoke access, and disconnected data is deleted within 30 days. Even so, handing a chatbot your medical history is a leap. Critics see where it leads.
“The interesting thing to me is the surveillance ratchet,” one wrote, imagining an AI that wants a camera, a microphone, and eventually your pheromones.
Awkward timing #
The launch could hardly be worse timed. Days earlier, a Florida pastor sued OpenAI, claiming ChatGPT gave him “extremely dangerous medical recommendations” that led him to delay care for a near-fatal pulmonary embolism.
OpenAI’s own disclaimer concedes the bot “can still make mistakes” and does not replace a doctor. Its heavy reliance on Apple Health is awkward, too, given OpenAI is fighting Apple in court over trade secrets.
None of this is slowing the march of AI into medicine, from note-taking to diagnosis. The upside is real. People juggling scattered records and overlapping conditions may finally see the whole picture.
The risk is just as real. A confident chatbot, trusted with the most sensitive data there is, is right most of the time and wrong when it matters most. OpenAI says its models keep improving. In the same breath, it says to check anything important with a real doctor.
Get the TNW newsletter #
Get the most important tech news in your inbox each week.