Apple on Sunday launched the Apple Health AI Usage Study, a U.S.-only study built around an iPhone app that answers everyday health questions. The study is open to up to 200,000 eligible participants age 21 and older. Apple says the work aims to inform the responsible design of health AI experiences for Apple and the industry.

The app combines medically grounded AI technology developed by OpenEvidence, a clinical AI platform widely used by healthcare professionals, with a research experience Apple designed. The company aims to run the study for more than 12 months.

What Changed

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Answers with citations

Participants can ask the study app everyday health questions. The answers draw on medical information from sources such as The New England Journal of Medicine, the Journal of the American Medical Association and Nature. Each response comes with linked, peer-reviewed clinical citations.

Full access to OpenEvidence is currently only available to health care professionals. Apple says it uses its own technology to turn that research into easy-to-understand answers. Users can rate the answers and complete surveys about their experience.

The app relies on text-based prompts, with dictation available through the iOS keyboard. It is not integrated with Siri or Apple Intelligence. Joining still requires an Apple Intelligence-enabled iPhone, meaning an iPhone 15 Pro or later, running iOS 26 or later.

Big AI labs, including OpenAI and Anthropic, already offer chatbots that answer health questions with specialized features. Earlier this year, OpenAI said users were asking over 300 million health-related queries each week.

The health data on the table

Participants can choose to share HealthKit data types such as resting heart rate, activity, sleep or lab results. Apple says researchers want to see whether personal health information makes answers more relevant and easier to understand. Someone asking how to improve their resting heart rate, for example, could receive advice based on their own health measurements rather than general recommendations.

Apple says questions pass through privacy protections designed to remove identifying information before reaching OpenEvidence. OpenEvidence deletes question-and-response data within 24 hours, according to Apple. Apple also says the data will be anonymized before it goes to a secure third-party cloud provider.

Study data will not be used to train AI models. Participants control whether they share any HealthKit metrics or survey responses. Apple says it has implemented security measures to protect the study data it maintains.

Where chatbots fall short

Apple's announcement says consumer health technologies powered by LLMs "may at times offer responses that are incomplete or unclear." The app has guardrails intended to detect certain concerning questions and suggest that a participant follow up with a provider or other resources. Apple says the app is not meant to diagnose or treat any condition, or replace the advice of medical professionals.

An Oxford study published in Nature Medicine in February 2026 found that AI chatbots accurately identify medical conditions nearly 95% of the time when tested on clinical cases. Everyday users reached the right answer less than 35% of the time when trying to understand dense guidance.

Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.

"Millions of people are turning to AI for health advice, but accurate answers aren't always useful. Confusing or incomplete information can create false reassurance or delay care," said Dr. John Brownstein, a professor of pediatrics and biomedical informatics at Harvard Medical School and an ABC News contributor.

What happens to the findings

"When people have access to clear, medically grounded information, they can walk into their doctor's appointments feeling more prepared, informed, and confident," Sumbul Desai, Apple's vice president of Health and Fitness, told ABC News.

"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, professor and chair of the Department of Medicine at UCSF, in Apple's announcement.

Apple told ABC News it plans to make the research available broadly, not only to inform its own future products. Apple has not announced whether the findings will lead to new AI-powered health features in its existing products.

"We need to understand not just whether AI provides accurate answers, but whether people understand those answers, trust them appropriately and ultimately make better health decisions," Brownstein said.

Frequently Asked Questions

Who can join Apple's Health AI Usage Study?

The study is open to up to 200,000 eligible participants age 21 and older in the United States. They need an Apple Intelligence-enabled iPhone, meaning an iPhone 15 Pro or later, running iOS 26 or later.

Where do the app's answers come from?

The app uses AI technology developed by OpenEvidence, a clinical AI platform widely used by healthcare professionals. Answers draw on medical information from sources such as The New England Journal of Medicine, the Journal of the American Medical Association and Nature, with linked, peer-reviewed clinical citations.

What happens to the questions participants ask?

Apple says questions pass through privacy protections designed to remove identifying information before reaching OpenEvidence. According to Apple, OpenEvidence deletes question-and-response data within 24 hours, and study data will not be used to train AI models.

Does the app use Siri or Apple Intelligence?

No. The app relies on text-based prompts, with dictation available through the iOS keyboard, and it is not integrated with Siri or Apple Intelligence.

Will the findings lead to new Apple health features?

Apple has not announced whether the findings will lead to new AI-powered health features in its existing products. Apple told ABC News it plans to make the research available broadly.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

OpenAI offers to pay future power bills; hacking AI rarely strays in tests
IMPLICATOR .ai Morning Briefing · From San Francisco   Friday, November 6, 2026 9 stops From San Francisco 1 The Editorial   Morning, humans. AI is getting a lesson
OpenAI pauses tool-use training; AWS traces why AI agents acted
IMPLICATOR .ai Morning Briefing · From San Francisco   Monday, September 28, 2026 11 stops From San Francisco 1 The Editorial   Marcus here. People still have to ch
Z.ai Delays GLM-5.3 Weights Two Weeks After Cyber Score Beats Mythos 5
Beijing-based Z.ai released GLM-5.3 on Friday while holding back the model's downloadable weights and gating its most sensitive cybersecurity functions. In launch tests, GLM-5.3 scored 84.5% on CyberG
AI News

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai