ChatGPT Can Now Read Epic Patient Charts for Doctors

OpenAI has plugged ChatGPT directly into Epic, the electronic health record system that holds the charts for a large share of American patients. Announced on 1 September, the integration lets clinicians pull authorised information from a patient's record and ask questions about it in plain language, without stitching the history together by hand.

Through ChatGPT for Healthcare, a doctor can ask what has changed since a patient's last visit, surface recent lab results, follow a medication history, or flag unresolved follow-ups. Epic's records reach an enormous population, and OpenAI says the tool draws on appointment notes, lab results, medications and specialist documentation across a chart.

Two design choices matter here. The access is read-only, so ChatGPT can pull from the record but cannot write anything back into it. And OpenAI paired the launch with a Healthcare Public Data plugin that links the workspace to nine official health datasets, so a clinician can check a patient question against population-level references in the same place.

How safe is it

OpenAI put some numbers behind the safety pitch. Across 4,363 ratings, physicians judged 99.1% of the model's responses safe across the tested use cases. That is a strong figure, and it is also a reminder that the remaining fraction lands on real charts, which is why the read-only limit and clinician sign-off are doing a lot of work.

The move fits a clear pattern. Americans are already bringing medical questions to chatbots on their own. A recent Pew survey found about a third now use AI for health information. Putting the same technology inside the clinician's workflow, with the actual chart attached, is a bigger step than a consumer asking a general chatbot about a symptom. It also raises the stakes on getting the guardrails right.