Apple launches health AI study with OpenEvidence

A translucent human figure made of layered medical diagrams stands in a quiet garden at dawn, surrounded by floating, dissolving text fragments from scientific journals; soft light filters through tre

Apple has launched a research initiative to understand how people use AI when managing personal health, marking a cautious but strategic step into the growing domain of medical chatbots. Rather than releasing a consumer AI health product, the company is running a controlled, longitudinal study in collaboration with OpenEvidence, a clinical AI platform trusted by healthcare professionals. The effort reflects Apple’s preference for evidence-based development, particularly in sensitive areas like health.

Who can join and how it works

The Apple Health AI Usage Study is open to US residents aged 21 and over who own an iPhone 15 Pro—not all iPhone 15 models—and are running iOS 26 or later. Although the app does not use Apple Intelligence, it requires devices capable of supporting the latest software. Participants install a dedicated iOS app to ask health-related questions in plain language. Responses are generated by OpenEvidence’s system, which draws exclusively from peer-reviewed clinical sources such as The New England Journal of Medicine, JAMA, and Nature. Answers are simplified for general understanding and include source references.

Users can rate responses for accuracy, detail, and relevance, and must complete monthly surveys. They may optionally share data from Apple’s HealthKit—such as activity levels, sleep patterns, resting heart rate, or lab results—to allow more context-aware answers. This integration aims to simulate how personalised AI health tools could function in real-world scenarios. The app supports text-based input and dictation via the iOS keyboard but is not integrated with Siri.

Privacy and data handling

Apple and OpenEvidence stress that privacy is central to the study’s design. All data shared by participants is anonymised before being transmitted to a secure third-party cloud provider. Apple explicitly states that the collected data will not be used to train any AI models—a notable distinction given industry practices elsewhere. OpenEvidence deletes individual query and response data within 24 hours, retaining only aggregated feedback for analysis.

Despite the use of clinical data sources, the app does not offer diagnoses or replace medical consultations. Instead, it guides users toward appropriate healthcare providers when questions suggest urgent or complex conditions. This aligns with Apple’s long-standing position of positioning its tools as supportive, not substitutive, in medical contexts.

Strategic positioning in the AI health race

While companies like OpenAI and Anthropic have rolled out general-purpose AI chatbots fielding millions of weekly health queries, Apple is choosing research over rapid deployment. The scale of public interest is clear: OpenAI reported over 300 million health-related prompts per week earlier in 2026. Yet Apple’s approach prioritises clinical validity and user comprehension over speed.

The study may inform future integrations into Siri, the Health app, or Apple Watch ecosystems. However, no such releases are confirmed. For now, Apple is gathering insights from up to 200,000 participants over at least 12 months—an approach that suits regulated markets like the UK and UAE, where digital health tools face stricter scrutiny.

What it means

For businesses and developers, Apple’s move signals that trust, clinical grounding, and transparency may become competitive advantages in health tech. In Kazakhstan, where digital health infrastructure is expanding, such models could offer a template for locally adapted, evidence-based AI tools. Founders should note that Apple is not betting on AI as a standalone product but as a feature validated through real-world use and ethical safeguards.

The study also underscores growing divergence in AI deployment strategies: while some firms prioritise scale and speed, others like Apple are investing in longitudinal research to reduce long-term risk. This could influence regulatory expectations, particularly in markets valuing data sovereignty and medical accountability.

Apple’s cautious stance offers a roadmap for responsible innovation in sensitive domains. The absence of AI training on user data, combined with short data retention and clinical source reliance, sets a high bar. Whether this translates into a public product remains to be seen—but the insights may shape health AI far beyond Cupertino.

Sources

Leave a Reply

Your email address will not be published. Required fields are marked *