Apple Health AI study: who can join
Apple opened a US research app on November 15, 2026 with OpenEvidence, for up to 200,000 adults 21 and older. It is a study, not a diagnosis.

Apple opened a US research app on November 15, 2026 with OpenEvidence, for up to 200,000 adults 21 and older. It is a study, not a diagnosis.

On November 15, 2026, Apple Newsroom ↗ announced the Apple Health AI Usage Study, a virtual study and an app, with OpenEvidence. It is open to up to 200,000 people, age 21 and older, in the United States. The phone on that page is an Apple Intelligence-enabled iPhone 15 Pro or later running iOS 26 or later.
The picture above is ours. We did not enroll. This page is not medical advice. The app is a research tool. Apple's footnote says it is not meant to diagnose or treat any condition, or to replace a medical professional, and that participants should consult a doctor or clinician about a health concern or to check medical information. Do that. Do not treat this article, or the study app, as a care plan.
The Verge ↗ (November 16) says the app does not use Apple Intelligence and is not integrated with Siri. Dictation, if used, is the iOS keyboard. The phone still has to be an Apple Intelligence model. TechCrunch ↗ wrote "iPhone 15 or later." Apple's newsroom says 15 Pro or later. The newsroom is the eligibility rule.
Fierce Healthcare ↗, quoting Apple, says people can remain for up to 12 months, on a study app separate from Apple Health and the Research app, overseen by an independent review board. TechCrunch wrote "for more than 12 months." The newsroom prints no length. Use Fierce's "up to 12 months."
This is not the paid add-on in our Apple Intelligence usage packs note. That product is a $49.99 Logic Pro pack after a monthly limit. This one is a study enrollment.
Apple says participants ask everyday health questions and see linked citations from sources such as The New England Journal of Medicine, JAMA, and Nature. HealthKit sharing is optional. The newsroom names resting heart rate, activity, sleep, and lab results. Fierce adds that labs enter only if the person has already imported clinical records into the Health app and agrees again to share them with this study.
Fierce reports that OpenEvidence's platform draws on licensed literature and guidelines through partnerships with those three journals plus Cochrane and the National Comprehensive Cancer Network. Apple's newsroom names the three journals. The extra two are Fierce's description of OpenEvidence, not a line on the newsroom page.
Fierce, attributing the path to Apple: a query is processed on the iPhone to remove identifying information, then goes to a secure study environment run by a third-party study contractor, where it is de-linked from the participant ID and screened again, then sent to OpenEvidence, which writes the answer from reference material. Apple says study data will not be used to train AI models. TechCrunch reports that Apple said OpenEvidence deletes the query and the response after 24 hours. Fierce's path does not mention that clock. The newsroom says study data will not train models, and that people control HealthKit sharing and survey answers. It does not, on its own, say whether OpenEvidence sees the question.
People rate the answers. Fierce says a baseline survey and monthly surveys. TechCrunch also says monthly surveys. The app has safety guardrails meant to detect certain concerning questions and point the person to a provider or other resources. Apple says the study will also look at preparing for a visit with the person's own doctor. That is a research question. It is not a reason to skip the visit.
A general chat is a different product. ChatGPT, Claude, and Gemini answer inside their own apps. This one, as Apple describes it, is a study client in front of OpenEvidence. Gemini Connected Apps is the other privacy fact on this site: human reviewers can see excerpts of connected Gmail and Drive. Read the consent screen before HealthKit goes on. We have not seen that screen.
"Apple spent a decade telling you your health data never leaves your phone. Now it wants up to 200,000 people to spend a year talking to a health chatbot." — BasicProtein ↗ (@BasicProtein26), November 16
"It doesn't say if OpenEvidence sees your questions or HealthKit data." — Aiholics ↗ (@Aiholics_), November 16. Fierce's account says the de-identified query is sent to OpenEvidence. HealthKit sharing is a separate optional step on Apple's page.
"Apple Health AI x @OpenEvidence research study!" — sarah guo ↗ (@saranormous), November 16
Reddit was not reachable for this story.
Enroll only as a research subject, consent text in front of you. On Apple's description, the draw is journal citations and a refusal to train on study data. The cost is health questions for up to 12 months, plus optional heart rate, sleep, activity, and labs leaving the phone after the on-device scrub Fierce describes. If that trade is unacceptable, do not enroll. The newsroom does not make the app a doctor.
Do not use it to choose a treatment, a drug, or whether to go to a clinic. Apple says the app is not for diagnosis or treatment. A citation is not a care plan. Concerning questions are supposed to be steered to a provider. Take that path, and talk to a clinician about your own health. This site will not tell you what the app's answer means for you.
Who should care. Adults in the US, 21 or older, on an iPhone 15 Pro or newer with iOS 26, who already wanted a cited health explainer and will read the consent. Who should skip it. Anyone who wanted answers that stay on the phone, anyone under 21, anyone outside the US, and anyone looking here for medical advice.