According to GSM Arena, OpenAI is launching a new feature called ChatGPT Health, a dedicated space within ChatGPT where users can connect accounts from Apple Health, MyFitnessPal, Function, Weight Watchers, AllTrails, Instacart, and Peloton. The company says it’s designed with physicians to help people manage their health, and it’s responding to the fact that over 230 million people globally already ask ChatGPT health questions. The Health section uses purpose-built encryption and isolation, and OpenAI promises not to use these chats for model training. For medical records, OpenAI is partnering with b.well, a secure US health data network. The feature can explain lab results, prepare appointment questions, and interpret wearable data. Right now, ChatGPT Health is available only to a small group of early users outside the European Economic Area, Switzerland, and the UK, with medical record integrations limited to the US.
The Privacy Play
Here’s the thing: the most interesting part of this announcement isn’t the features—it’s the privacy promises. OpenAI knows health data is the ultimate third rail. So they’re leading with “purpose-built encryption,” data isolation, and a firm “we won’t train on this” pledge. That’s a direct response to the massive trust deficit they and other AI companies face. But let’s be real. The real test won’t be the press release; it’ll be the first security audit or, heaven forbid, a data incident. Can they actually compartmentalize this data when their entire business is built on ingesting information to make models smarter? They’re betting their reputation on it.
Stakeholder Shifts
For users, this is a classic convenience-vs.-privacy trade-off. Getting a plain-English explanation of your weird lab results is incredibly appealing. But you’re handing over your most sensitive data to a company whose core competency is language models, not HIPAA compliance. For developers and health apps, this is a double-edged sword. Integration with a giant like ChatGPT could drive user engagement for services like MyFitnessPal or Peloton. But it also potentially makes them mere data feeders to a more powerful, centralized AI interface. Why open your app when you can just ask ChatGPT? And for the healthcare market, it’s a slow but steady encroachment. AI is moving from the back office into the direct patient-facing role, interpreting data and guiding conversations. That’s going to rattle some cages.
The Big Picture
Basically, OpenAI is formalizing a behavior that’s already happening. 230 million health queries is a staggering number they just casually dropped. They’re not creating a new market; they’re trying to corral an existing, risky one into a safer, monetizable pen. The partnership with b.well for medical records is the tell. They’re not just about step counts and calories anymore—they’re aiming for the full clinical picture. So, what’s the endgame? It’s hard to see this as anything but a foundational step. First, it’s a wellness assistant. Next, it could be a triage tool, a chronic condition coach, or an integrated part of electronic health records. They’re building the data pipeline and user trust, one de-identified chat at a time. The rollout is cautious now, but the ambition is clearly anything but.
