Apple launches new app to study how people use AI for health

AI & Compute

Apple Recruits Up to 200,000 iPhone Users for Health AI Study

Apple launched a study app with OpenEvidence to test clinically grounded health AI answers with up to 200,000 U.S. iPhone users over more than 12 months.

By
Rebecca Stone
Filed
Channel
AI & Compute
Read
3 min read

Apple wants up to 200,000 people to help it figure out whether AI can safely answer medical questions. The company launched a new study app today, built with healthcare-focused AI company OpenEvidence, that invites U.S. iPhone users over 21 to ask health questions and get answers grounded in clinical literature rather than the open web.

It's a deliberately cautious move. OpenAI and Anthropic have already shipped general-purpose chatbots that field health queries; Apple hasn't. Instead, the iPhone maker is running a formal research study — more than 12 months long, capped at 200,000 participants — before it commits to putting medical AI into the Health app or Siri. Whether it does remains unclear; Apple hasn't said.

Who can join, and what they get

Eligibility is narrow and precisely defined:

  • U.S. residents over 21
  • iPhone 15 or later
  • Running iOS 26 or later with Apple Intelligence enabled

The app itself works like a chatbot. Users type health questions, much as they would into ChatGPT or Claude. The difference is the grounding: answers draw on clinical sources including The New England Journal of Medicine, the Journal of the American Medical Association and Nature. Apple said the app will aim to simplify those sources into plain answers for everyday medical questions — not diagnostics.

Participants rate the answers and give feedback on each response. They'll also complete monthly surveys. The stakes for Apple are commercial as much as scientific. OpenAI said earlier this year that its users ask more than 300 million health-related queries every week — a volume that suggests enormous latent demand Apple has so far left to competitors.

How Apple handles the data

The study pulls in more than typed questions. Participants can share HealthKit data — resting heart rate, activity, sleep, lab results — so the app can return more personalized, pointed suggestions. Apple also framed a practical use case: preparing people for doctor's visits, and guiding users toward specific providers when a question touches on concerning medical issues.

Privacy terms are unusually explicit for an AI product. Apple said all data will be anonymized before it reaches a secure third-party cloud provider, and that none of it will be used to train AI models. OpenEvidence, for its part, deletes query and response data after 24 hours, according to Apple.

That distinction matters. Rival AI labs routinely use user interactions to improve their models. Apple is explicitly walling this study off from its training pipeline — a positioning choice that fits its long-standing privacy marketing, and one it will likely lean on if the technology graduates from study to product.

What Apple's executives are saying

Dr. Sumbul Desai, Apple's vice president of Health and Fitness, framed the study as a matter of scientific method rather than speed to market.

"At Apple, we believe responsible innovation starts with scientific rigor and clinical grounding," Desai said in a statement. "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."

She added: "Insights from this study will help advance the broader scientific conversation and ensure that as technology evolves, we continue building trusted tools that genuinely support people wherever they are on their health journey."

Study first, product later

The app is available to download now for eligible U.S. users. Everything beyond that — an AI answering feature inside the Health app, a medically informed Siri, some other delivery mechanism — stays on the roadmap, undefined and unconfirmed. What Apple has committed to is only the study itself: 200,000 people, twelve-plus months, clinically grounded answers, and a feedback loop designed to tell the company what users actually trust.

By the time results land, the market will have moved again. OpenAI's 300 million weekly health queries show the demand is already there; Apple's bet is that clinical grounding — verifiable sourcing, anonymized data, no model training — will prove the more durable differentiator in a category where being wrong carries real consequences.

Source: techcrunch.com

Share this article:

More from Rebecca Stone

Rebecca Stone

Show full bio

Correspondent covering technology and product news in Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.

1,037 articles

Related articles

« Previous articleNext article »