AI in medicine / The AI briefing

Apple starts a 200,000-person US study of how people understand and trust AI health answers

Apple's new study app, built with clinical AI company OpenEvidence, answers everyday health questions with linked sources and asks users to rate the replies. It runs only in the US, and Apple says the data will not train AI models. What it shows about building AI answers people can rely on.

Two hands holding an iPhone in a clear case at a table, next to a takeaway coffee cup
A person using an iPhone. Illustrative stock photo, not of the Apple study app. Photo: Kristin Hardwick / ISO Republic (via Wikimedia Commons), licence CC0 1.0; cropped.

Apple announced the Apple Health AI Usage Study on 15 November. Participants in the US use a dedicated app to ask everyday health questions and receive answers generated by OpenEvidence, a clinical AI platform used by healthcare professionals, with links to peer-reviewed sources. Apple is looking for up to 200,000 participants aged 21 and over with an iPhone 15 Pro or later, and says the study will examine trust, the quality of responses, comprehension and what people do with the answers. Apple says study data will not be used to train AI models.

What Apple launched

Apple's announcement describes a virtual research study and app "designed to advance the understanding of how people interact with conversational health AI tools." Participants can ask everyday health questions, read answers written in approachable language and follow linked, peer-reviewed clinical citations. OpenEvidence generates the answers using medical information from sources such as The New England Journal of Medicine, the Journal of the American Medical Association and Nature.

The study is open to up to 200,000 eligible people aged 21 and older across the US who have an Apple Intelligence-enabled iPhone (iPhone 15 Pro or later) running iOS 26 or later. According to Fierce Healthcare, Apple said participants may stay in the study for up to 12 months and will receive baseline and monthly surveys.

Participants can choose to share HealthKit data such as resting heart rate, activity, sleep or lab results, so the study can test whether personal data makes answers more relevant and easier to understand. Apple says the study will also explore whether the app helps people prepare for visits to their own doctors.

What the study measures

Apple says it will examine "what drives user trust, the quality of responses, comprehension, and everyday health actions", to inform the design of health AI for Apple and the wider industry. Participants rate the answers; The Verge lists accuracy, detail and relevance among the criteria. Fierce Healthcare reports that participants are asked for demographic details such as age, race and education, so researchers can see how these affect preferences and communication needs, and that Apple describes the study as overseen by an independent review board.

Apple is open about why it is testing first. Its announcement says consumer health tools powered by large language models "may at times offer responses that are incomplete or unclear, which is why research is critical to presenting health information to users with greater clarity."

Demand is already there. Fierce Healthcare cites a KFF survey finding that roughly one in three US adults have turned to an AI chatbot for health information.

How privacy and safety are handled

Apple says study data will not be used to train AI models and that participants decide whether to share HealthKit metrics or survey answers. Fierce Healthcare reports, citing Apple, that questions are first processed on the iPhone to remove identifying information, then pass to a secure study environment run by a third-party contractor, where they are de-linked from participant IDs and screened again before reaching OpenEvidence. TechCrunch reports that OpenEvidence deletes query and response data after 24 hours.

The app also has safety guardrails intended to detect certain concerning questions and suggest that the participant follows up with a provider or other resources.

Apple says "The study app is not meant to diagnose or treat any condition, or replace the advice of medical professionals."

The study is US-only, and Apple has not announced a UK or EU version. TechCrunch notes it is not clear whether Apple plans AI answers for its Health app or Siri, and The Verge reports that the study app is not integrated with Siri or Apple Intelligence.

What businesses can learn from it

Evoogen's reading: few businesses will build a health assistant, but many already put AI answers in front of customers, from booking questions to product advice. Apple's design shows what matters when an answer carries risk: base answers on named sources and show them, check whether people understood the reply rather than only whether it is correct, point a person who asks a concerning question toward a clinician or other help, and keep data only as long as needed.

Apple is spending up to a year and up to 200,000 users on testing. A small business cannot run a study on that scale, but it can read a sample of real AI replies each week, ask some customers whether the answer helped, and set a clear rule for when the assistant must hand over to staff.

If you want a customer-facing AI assistant that answers from your own information, hands sensitive questions to staff and keeps data no longer than needed, Evoogen can design and set it up. Talk to Evoogen about a trustworthy AI assistant.

Sources

Evoogen checked these sources on 16 November 2026. Company and research claims stay attributed to their source; Evoogen analysis is labelled.

  1. TechCrunch: Apple launches new app to study how people use AI for health (15 November 2026, US time) — checked 2026-11-16
  2. Apple Newsroom: Apple launches research study to explore the use of AI for health (15 November 2026) — checked 2026-11-16
  3. The Verge: Apple’s new study looks to improve the effectiveness of AI health tools (15 November 2026, US time) — checked 2026-11-16
  4. Fierce Healthcare: HLTH26: Apple kicks off 200K-person health AI study (November 2026) — checked 2026-11-16