200,000 iPhone owners will ask an AI their health questions — and grade every answer
Apple's Health AI Usage Study, conducted with OpenEvidence, will recruit up to 200,000 adults across the United States to ask a conversational AI their everyday health questions — answers drawn from The New England Journal of Medicine, JAMA and Nature — while Apple measures whether they trust what it says, whether they understand it, and what they do next.

Photo: Miriam Alonso, via Pexels (Pexels License — free to use). A woman in bed typing on her smartphone — the everyday act the study will measure; the image illustrates that act, not the study app itself.
The scene is already ordinary: a symptom you cannot name, a lab result you cannot read, a question you would rather not ask out loud — typed into a phone, answered by a machine. What Apple announced on November 15 is an attempt to move that scene out of the shadows of everyday use and into a laboratory. The Apple Health AI Usage Study is a virtual research study, with its own app, and it is looking for up to 200,000 volunteers in the United States willing to do inside a study what millions already do outside one: ask an AI about their health.
The study is conducted in collaboration with OpenEvidence, and its object is not the AI's knowledge — it is the people using it. According to Apple, the study will examine what drives user trust, the quality of the responses, how well participants comprehend them, and the everyday health actions that follow. The stated aim is to «inform the responsible design of health AI experiences» — Apple's, and the industry's.
An AI that answers from the journals
The answers inside the study app do not come from a general-purpose chatbot. They are generated with technology from OpenEvidence, a clinical AI platform that Apple describes as widely used by healthcare professionals, and that builds its answers from established medical sources — Apple names The New England Journal of Medicine, the Journal of the American Medical Association and Nature. Each response arrives with linked, peer-reviewed clinical citations, so the participant can see the paper standing behind the paragraph.
That citation layer is the study's central bet. Consumer health tools powered by large language models «may at times offer responses that are incomplete or unclear», Apple concedes in its own announcement — which is why, the company argues, research is critical. The study app is designed to communicate complex medical information in approachable language, and one of the things being measured is precisely whether the receipts change the reader: whether an answer with sources earns more trust, and better understanding, than an answer without them.
Participants will be able to ask everyday health questions, and to rate what comes back. TechCrunch reported that the app asks users to evaluate the responses, and that participants will complete monthly surveys across a study planned to run for more than 12 months. Apple has also built in a practical purpose: the study will explore how health AI can help participants better prepare for clinical visits with their own doctors — arriving at the appointment with better questions, rather than with a self-diagnosis.
Who can join — and what joining means
Enrollment is open to up to 200,000 eligible participants aged 21 and older across the United States. The hardware bar is Apple's own ecosystem at its newest end: an Apple Intelligence-enabled iPhone — iPhone 15 Pro or later — running iOS 26 or later. The study app is free to download, and enrollment happens inside it.
What a participant hands over starts with the questions themselves: the symptoms, worries and lab results they type into the app over the course of a year or more. On top of that, participants can choose to share HealthKit data — Apple names resting heart rate, activity, sleep and lab results — to explore how personal data integration changes the relevance and the understanding of the answers. Apple stresses the choice architecture: participants retain control over whether to share any HealthKit metrics or survey responses.
The privacy terms are part of the announcement, and they are specific. The study, Apple says, «was designed from the ground up with privacy protections, transparency, and clinical guardrails, and study data will not be used to train AI models». Apple told TechCrunch that the data is anonymized before it is sent to a secure third-party cloud provider, and that OpenEvidence deletes the question and answer data after 24 hours. The fine print of what each participant accepts lives in the study's informed consent — the document a volunteer reads, or scrolls past, before the first question is asked.
One clause deserves a slow read. Under Apple's privacy policy for its health study apps, withdrawing from a study is not the same as erasing it: Apple states that it may not delete the information it has already collected, and may continue to use it. Joining is a tap; leaving is a tap; unwinding what was shared in between is a different matter.
The limits are printed on the box
Apple's own footnote is unambiguous: «The study app is not meant to diagnose or treat any condition, or replace the advice of medical professionals.» Participants are told to consult a doctor or clinician to address health concerns or verify medical information. The app also carries safety guardrails intended to detect certain concerning questions and to suggest that the participant follow up with a provider or other resources if needed.
Those limits define what is actually being tested. The study does not ask whether an AI can be a doctor. It asks something prior, and in commercial terms more valuable: when an AI answers health questions with clinical sources, in plain language, with citations attached — do people trust it, do they understand it, and do they act on it? And where does that trust break?
Why Apple is running a study instead of shipping a product
«At Apple, we believe responsible innovation starts with scientific rigor and clinical grounding», said Sumbul Desai, M.D., Apple's vice president of Health and Fitness, in the announcement. «Health is deeply personal, and conducting a dedicated research study is the most thoughtful way to understand how people engage with, comprehend, and evaluate health AI in everyday life.» The insights, she said, will help advance the broader scientific conversation about trusted health tools.
The academic voice in the announcement makes the same point from the other side of the screen. «Patients and clinicians are both turning to AI, but the tools they use are being built and tested in separate worlds», said Robert M. Wachter, M.D., professor and chair of the Department of Medicine at UCSF. «When patients and doctors draw on versions of the same trustworthy tools, patients arrive prepared, conversations improve, and next steps happen seamlessly.» OpenEvidence is, notably, a tool clinicians already use — which means the study puts patients in front of a version of the same instrument their doctors may be consulting.
What Apple has not announced is a product. There is no date for a health chatbot inside the Health app or Siri, and no published timetable for the study's results. What exists, as of this week, is the recruitment: 200,000 seats, a free app, and a year of questions.
The questions are already being asked
Apple is not creating the behavior it wants to study; it is formalizing it. Covering the launch, TechCrunch recalled OpenAI's earlier disclosure that its users ask more than 300 million health-related questions a week. The traffic already exists, at enormous scale, inside general-purpose chatbots that carry no clinical sourcing requirement and no consent form. The study's premise is that the same questions, asked inside a designed environment — clinical sources, linked citations, safety guardrails, a signed consent — can actually be measured: what people ask, what they are told, whether they believe it and what they do with it. That measurement, Apple is betting, is worth more right now than another product launch.
The decision in front of the reader
Strip away the research language and the offer is simple. In exchange for answers drawn from the medical literature — with citations, in plain language, on the phone already in your pocket — the study asks for the most intimate dataset a person produces: what they ask about their own body when nobody is listening. Apple has published its terms for that exchange: anonymized data, no training use, questions deleted by OpenEvidence within 24 hours, HealthKit sharing under the participant's control, and a consent form that carries the details.
Whether those terms are enough is, in a sense, the study's first result — it will be visible in how many of the 200,000 seats fill, and how fast. The answers will take more than a year to read. The questions start now.
𝐆𝐄𝐓 𝐅𝐑𝐄𝐄 𝐅𝐔𝐋𝐋 𝐌𝐀𝐆𝐀𝐙𝐈𝐍𝐄 𝐀𝐓 thebuyerspulse.com
Sources
- Apple Newsroom, «Apple launches research study to explore the use of AI for health» (November 15, 2026) — the study's announcement: collaboration with OpenEvidence, what it will evaluate (trust, response quality, comprehension, everyday health actions), the medical sources (NEJM, JAMA, Nature) and linked peer-reviewed citations, optional HealthKit data types (resting heart rate, activity, sleep, lab results), the privacy design and the no-training commitment, the safety guardrails, eligibility (up to 200,000 participants, age 21+, iPhone 15 Pro or later, iOS 26 or later), the statements by Sumbul Desai and Robert M. Wachter, and the footnote that the app does not diagnose or treat.
- TechCrunch, «Apple launches new app to study how people use AI for health» (November 15, 2026) — the study's planned duration of more than 12 months with monthly surveys and response ratings; the anonymization of data before it goes to a secure third-party cloud provider, and OpenEvidence's deletion of question and answer data after 24 hours, as stated by Apple; and OpenAI's earlier disclosure, recalled in the coverage, of more than 300 million health-related questions a week from its users.
- Apple, Health Study Apps Privacy Policy — withdrawal from a study does not necessarily delete information already collected, which Apple states it may retain and continue to use.
- Photo: Miriam Alonso, via Pexels (Pexels License — free to use). A woman in bed typing on her smartphone; illustrative of the everyday act, not the study app.