Clinicians at participating health systems can now bring patient records from Epic, the widely used electronic health record system, into ChatGPT for Healthcare, OpenAI announced on Sept. 1. Authorized clinicians can pull in appointment notes, lab results, medications, and specialist documentation, then ask the AI to summarize a patient's history or flag what has changed since the last visit.
For patients, the shift is mostly invisible but potentially significant. The clinician preparing for your next appointment may be reading an AI-generated summary of your chart instead of clicking through it tab by tab. That could save time, but it also raises fair questions about accuracy, privacy, and what gets left out.
The potential reach is large. Epic holds records for more than 325 million patients, and OpenAI said the integration is read-only, meaning the AI cannot write anything back into a patient's record, TechCrunch reported.
Two Ways the Epic Connection Works
According to OpenAI's announcement, the integration works in two ways. In the first, clinicians bring authorized patient information into a ChatGPT workspace to review history, identify changes, and prepare for visits. In the second, available in some deployments, ChatGPT is built directly into the Epic screen so clinicians do not have to leave the chart.
OpenAI said clinicians can ask, for example, which recent lab results to review before an appointment, and that the AI points back to supporting information in the chart.
Access is limited to organizations, not individual clinicians. Organizations need a business associate agreement, the contract HIPAA requires when an outside company handles protected health information. OpenAI lists role-based access controls, single sign-on, and audit logs among its safeguards.
UCSF Is the Named Pilot Partner
UCSF Health in San Francisco is the only health system OpenAI named as a pilot partner for the Epic integration. "As a pilot partner, we're exploring how the new EHR integration with ChatGPT for Healthcare can help clinical teams understand what has changed and what matters most across a complex patient record," said Suresh Gunasekaran, president and CEO of UCSF Health, in OpenAI's announcement. He added that UCSF is engaging frontline teams to validate the capabilities in practice.
OpenAI also lists AdventHealth, Baylor Scott and White Health, Boston Children's Hospital, Cedars-Sinai, HCA Healthcare, Memorial Sloan Kettering Cancer Center, and UCSF as launch partners for ChatGPT for Healthcare more broadly. It has not said which of them, beyond UCSF, use the Epic connection.
Epic has issued no statement, a health system CIO reported, leaving open how closely the two companies are working together. Epic already sells its own AI tools for chart review and summarization.
Accuracy, Privacy, and the Evidence Gap
OpenAI said physicians rated 99.1% of responses safe across 4,363 ratings in 27 clinical use cases, such as pre-visit review and medication review. Those figures come from the company's own evaluation. No methodology has been published, and the results have not been independently replicated or peer-reviewed. The remaining 0.9% works out to about 39 ratings that fell short, and a single missed allergy or medication change can matter a great deal to one patient.
OpenAI has maintained that AI is not suitable for diagnosis or treatment decisions, which remain the clinician's responsibility. The company also faces lawsuits over health advice given through consumer ChatGPT, a separate product from the enterprise tool described here.
There is no independent data yet on whether AI chart summaries like these reduce errors, save meaningful time, or improve patient outcomes.
Patients do not need to take any action, and there is no cost to them. It is reasonable to ask whether your health system uses AI tools to summarize your chart, whether a clinician reviews those summaries, and how your data is protected.
Under the HIPAA Privacy Rule, patients have a right to access their health records and to request corrections. Reviewing your own chart through a patient portal is one of the most practical ways to catch errors, since any AI summary can only be as accurate as the record it reads.
Older adults, people with complex conditions, and patients who see many specialists have the most to gain from better chart summaries. They also have the most to lose if a summary misses something buried in a long record. For now, AI may be helping your doctor prepare, but your own questions and your own review of your record still count.
Key Questions Answered
What changed with Epic and ChatGPT? Health systems can now connect Epic records to ChatGPT for Healthcare, letting authorized clinicians summarize notes, labs, and medications. OpenAI announced the change on Sept. 1.
Can ChatGPT change my medical record? OpenAI says the integration is read-only and does not write information back into the record.
Which hospitals are using it? UCSF Health is the named pilot partner. OpenAI has not said which other health systems use the Epic connection.
Is my data protected? Organizations must have a HIPAA business associate agreement, and OpenAI lists access controls and audit logs among its safeguards.
How accurate is it? OpenAI reports physicians rated 99.1% of responses safe in its own testing. That data has not been independently reviewed.
What can patients do? Ask whether AI is used to prepare for your visits, and review your records through your patient portal.
