Key Points
- Apple launches research study with 200,000 US participants to examine health AI usage
- App uses clinical data from New England Journal of Medicine and JAMA for responses
- Personal data anonymised before cloud transmission and deleted after 24 hours
Apple has launched a new research application to study how people use artificial intelligence, software that processes information and generates responses based on patterns in data, for health-related questions, partnering with healthcare AI company OpenEvidence to test whether AI answers backed by clinical data can safely assist users with medical queries.
The study, called the Apple Health AI Usage Study, will enrol up to 200,000 participants in the United States over a period exceeding 12 months, according to the company. The launch comes as AI chatbots face increasing use for health questions, with OpenAI recently disclosing that users ask its chatbots more than 300 million health-related questions every week.
Participants must be 21 years or older, own an iPhone 15 or newer model running iOS 26 or later, and have Apple Intelligence enabled on their device, the company said. The app allows users to type health questions and receive responses drawn from clinical data published in medical journals including The New England Journal of Medicine, JAMA and Nature.
Users can optionally connect the app to their Apple Health data through HealthKit, the company’s health data framework. This enables the app to access information such as sleep patterns, resting heart rate, daily activity levels and laboratory results to provide more personalised responses, Apple said. The company added that the app can help users prepare for medical appointments and direct them to healthcare providers for concerning issues.
Privacy safeguards
Apple said personal data collected through the app will not be used to train any AI models. Information is anonymised before transmission to a secure third-party cloud provider, meaning it cannot be linked to individual users, the company stated. OpenEvidence deletes all user questions and AI response data after 24 hours, according to Apple.
Participants can rate the answers they receive, provide feedback on response quality and will be asked to complete monthly surveys as part of the research, the company said.
Sumbul Desai, Apple’s vice president for health and fitness, said the company believes responsible innovation starts with science and clinical grounding. “Health is deeply personal, and conducting a dedicated research study is the most thoughtful way to see how people engage with and evaluate health AI in their everyday lives,” Desai said.
The insights from the study will help Apple build trusted tools that support people on their health journey, the company said. Apple has not disclosed whether it plans to integrate the AI health response feature into its main Health app or Siri voice assistant.
The study positions Apple differently from companies such as OpenAI and Anthropic, which already offer general-purpose chatbots that respond to health queries. Apple’s approach focuses specifically on testing clinically grounded responses with oversight from medical literature sources rather than deploying a general conversational AI for health purposes.
By the numbers
Key figures from this story- 300 million
- Weekly health questions asked to OpenAI chatbots
- 200,000
- Target participants for Apple study
- 24 hours
- Data retention period before deletion
The research comes amid growing regulatory and medical community scrutiny of AI systems providing health information. While AI chatbots have demonstrated capability in answering certain medical questions, concerns persist about accuracy, liability and the potential for users to rely on AI responses instead of consulting healthcare professionals for serious conditions.
Your Questions, Answered
Who can participate in Apple's health AI research study?
The study is currently limited to users in the United States who are 21 years or older, own an iPhone 15 or newer model, and have iOS 26 or later with Apple Intelligence enabled.
How does Apple protect user privacy in the health AI app?
Apple anonymises all personal data before transmission to cloud providers, does not use collected data for AI training, and OpenEvidence deletes all questions and responses after 24 hours.
What medical sources inform the AI health responses?
The app draws on clinical data from peer-reviewed medical journals including The New England Journal of Medicine, JAMA and Nature to generate health-related answers.
How long will Apple's health AI study run?
The study will run for more than 12 months and aims to enrol up to 200,000 participants in the United States.




