Be honest. When was the last time you asked a chatbot about a weird pain or a confusing test result? If you have, you’re far from alone. And now Apple wants to know exactly how that goes.
On Sunday, Apple announced the Apple Health AI study, a research project built with OpenEvidence, a clinical AI tool that doctors already use. Up to 200,000 people in the US can sign up. They’ll ask health questions in an app and get plain-language answers backed by medical research. Then Apple will look at what they trust, what they understand and what they do next.
So what did Apple actually launch?
It’s a virtual study with its own app, called the Apple Health AI Study App. You download it, join, and start asking everyday questions. For example, you might ask what a lab result means or how to sleep better.
The answers come from OpenEvidence. It’s a platform built for healthcare professionals, and it draws on sources like the New England Journal of Medicine, JAMA and Nature. Each answer links back to peer-reviewed research, so you can check where a claim came from.
OpenEvidence isn’t new to big partnerships, either. In September, it teamed up with Anthropic to offer free medical AI to doctors. So Apple picked a partner with some clinical credibility, which makes sense for a company this careful about health.
There’s also a feature that helps you prepare for a doctor’s visit. Apple wants to see whether people walk into appointments better informed. Honestly, that’s the part we’re most curious about.
How the Apple Health AI study works

Who can join the Apple Health AI study
You need to live in the US and be 21 or older. You also need an iPhone 15 Pro or later running iOS 26, since those phones support Apple Intelligence, according to iClarified. So if you’re reading this in Yerevan, you can’t sign up. Still, what Apple learns here could shape the health tools you’ll use later.
What happens with your data
You can choose to share HealthKit data, such as resting heart rate, activity, sleep and lab results. Apple will test whether that makes answers more relevant and easier to follow. You can also fill out short surveys rating the answers you get.
Apple says it won’t use study data to train AI models. Sharing is optional, too. The app also has guardrails that spot certain worrying questions and nudge you to contact a doctor instead.
The App Store listing hints at a few more uses. It mentions putting your personal health data in context and building a personalized exercise or meal plan. That sounds handy. But it’s also exactly where good privacy rules matter most.
Why this one’s a bigger deal than it looks
Apple isn’t first to health AI. OpenAI launched ChatGPT Health in January. Then Microsoft and Perplexity followed in March with their own health tools. So why does a study matter more than another product launch?
Because nobody really knows how regular people use these answers. Robert Wachter, who chairs the Department of Medicine at UCSF, put it well, as quoted by The Next Web: “Patients and clinicians are both turning to AI, but the tools they use are being built and tested in separate worlds.”
In other words, doctors get tools tested in clinics, while patients get whatever’s on their phone. A study with up to 200,000 people could start to close that gap. Apple’s VP of Health and Fitness, Sumbul Desai, framed it this way: “At Apple, we believe responsible innovation starts with scientific rigor and clinical grounding.”
For now, the app isn’t tied into Siri or Apple Intelligence. Our guess? If the study goes well, that won’t last forever.
Okay, but should you trust a health chatbot?
Here’s our honest answer: trust it like a well-read friend, not like your doctor. Apple says as much itself. Large language models “can provide incomplete or unclear responses,” it notes, and the app isn’t meant to diagnose or treat anything.
That’s refreshing, frankly. Plenty of AI tools bury that warning in fine print. Here it’s front and center.
So use these tools to get your bearings. Ask what a term means. Write down questions before an appointment. But when something feels off, call a real person. We made a similar point about letting an AI read your bank account: convenience is great, though you should stay in charge.
Here’s a quick checklist we’d use. First, does the answer cite a real source? Next, does it tell you when to see someone? Finally, does it admit what it doesn’t know? If the answer is no to all three, close the app and call your clinic instead.
What the Apple Health AI study could change
For us at EIF, the big lesson isn’t really about Apple. It’s about testing. AI answers already pop up when you search, as we saw with Google’s newest model landing in Search. Yet we still know very little about how people act on those answers.
The Apple Health AI study is one of the first big attempts to find out. If you build health or AI products in Armenia, watch what it measures: trust, accuracy, understanding and next steps. Those are the right questions for any team, big or small.
And here’s the open question for all of us. When the AI says “probably nothing to worry about,” will we still go see a doctor?

