OpenAI just connected ChatGPT Health to Epic, the electronic health record system that holds data for more than 325 million patients. According to TechCrunch AI, the integration lets clinicians pull patient data into ChatGPT and ask questions in plain language, and in some setups it drops ChatGPT directly into the EHR workflow so doctors never leave a patient chart.
This matters because Epic is the backbone of American healthcare data. Wiring a general-purpose AI into it moves ChatGPT from a consumer curiosity to a working tool inside the clinic. Here’s what OpenAI actually shipped.
- Read access to the full patient picture. Clinicians can now surface appointment notes, lab results, medications, and specialist documentation, then have ChatGPT summarize them. TechCrunch AI reports they can also scan patient history, spot changes over time, and prep for upcoming visits.
- In-workflow pre-visit review. In some deployments, doctors can run a pre-visit review and build clinical timelines without leaving the patient chart. That’s the difference between a tool you switch to and one that lives where you already work.
- Read-only by design. OpenAI was specific on this point: the integration only reads health records. The AI doesn’t write anything back into Epic. That’s a deliberate guardrail, and a smart one given the stakes.
- A new Healthcare Public Data plug-in. Beyond patient charts, OpenAI added a plug-in that pulls from ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. The company says it helps staff synthesize trial eligibility criteria, medication identifiers, coverage policy versions, and provider records, the kind of cross-referencing that normally eats hours.
- Compliant workspace tools. Organizations with a Business Associate Agreement can now use ChatGPT Work, Codex, apps, and connectors inside their workspace for compliant workflows. That’s OpenAI signaling it wants to be the platform hospitals build on, not just a chatbot they tolerate.
Why the timing is loud
Last month OpenAI rolled out ChatGPT for health to all U.S. consumers, and it says people now fire 300 million health-related queries at ChatGPT every week. The Epic deal is the professional-grade companion to that consumer push. The company is clearly betting healthcare is one of AI’s biggest markets.
But OpenAI keeps drawing one hard line. Despite these deep integrations, it maintains that AI is not suitable for diagnosis or treatment. The Epic tool is built for synthesis and prep, not clinical decisions.
The safety question is real
OpenAI ran the numbers to make its case. Per TechCrunch AI, it collected more than 4,300 responses from physicians across 27 clinical use cases, including pre-visit review, clinical timelines, medication review, and handoff summaries, and found 99.1% of responses were safe.
Here’s the catch worth sitting with: in medicine, even a few unsafe answers can cause real harm. And the legal backdrop isn’t quiet. A few days before this rollout, a Florida pastor sued the company, alleging ChatGPT gave him a near-fatal recommendation. Back in May, family members of a user sued over what they said was wrongful dosage advice. Those cases are the shadow hanging over every healthcare AI launch right now.
What stands out
OpenAI is threading a careful needle. It’s pushing ChatGPT as deep into the medical system as it can while insisting the tool doesn’t diagnose or treat. The read-only restriction and the BAA compliance track show a company that has learned to move into regulated territory with lawyers in the room.
The open question is adoption. Clinicians are cautious by training, and a 99.1% safety rate reads differently to an engineer than to a physician holding a patient’s chart. Whether doctors trust an AI summary enough to lean on it during a real visit is the test that matters, and no press release settles it.
For now, the direction is clear. AI is moving from the waiting room to the exam room, one read-only integration at a time. Full details are available at the original TechCrunch AI report.