OpenAI is taking another step into healthcare with a new integration between ChatGPT for Healthcare and Epic, one of the major electronic health record (EHR) systems used by healthcare organisations. The company announced the new integration on September 1, 2026. It allows authorised healthcare teams to bring patient information from supported Epic environments into ChatGPT for Healthcare and use AI to review and summarise the information. What can clinicians do with the integration? Medical records can contain a large amount of information spread across different sections of a patient's file. Doctors may need to go through previous appointment notes, laboratory reports, medication information and specialist documentation before meeting a patient. OpenAI's new integration is designed to make this process easier. Clinicians can ask questions about a patient's medical history, including what has changed since the previous visit, which recent laboratory results need attention, whether there have been medication changes and what follow-ups or unresolved issues are present in the record. ChatGPT can then bring together relevant information from the authorised patient record and provide a summary while pointing back to the supporting information in the chart. ChatGPT can also work inside the EHR The integration has two main approaches. In the first, authorised patient information from an EHR can be brought into ChatGPT for Healthcare so clinicians can review the patient's history and prepare for appointments. In supported healthcare deployments, ChatGPT can also be integrated directly into the EHR interface. This means clinicians can use AI-assisted workflows without having to leave the patient chart. This could be particularly useful for tasks such as preparing for a consultation, reviewing a patient's medical timeline and quickly understanding changes in a long record. It is read-only One important point is that the Epic integration is read-only. ChatGPT can access and analyse authorised information, but it does not automatically write its generated information back into the patient's medical record. This provides an additional separation between AI-generated output and the official clinical record. Clinicians still remain responsible for reviewing the information and making medical decisions. UCSF Health is testing the technology UCSF Health is participating as a pilot partner for the new integration. The organisation is exploring how the technology can help clinical teams understand what has changed and what is most important across complex patient records. The aim is not simply to add another AI tool for doctors. Instead, OpenAI is trying to connect AI with the information healthcare teams already use during their daily work. OpenAI is also adding public healthcare data Along with the Epic integration, OpenAI has introduced a Healthcare Public Data plugin for ChatGPT for Healthcare. The plugin provides structured access to nine official healthcare data sources, including PubMed, ClinicalTrials.gov, DailyMed, RxNorm and CMS Coverage. This can help healthcare teams work with information such as clinical trial eligibility requirements, medication identifiers, coverage policies and provider records without having to search each source separately. Privacy and security remain important Healthcare data is highly sensitive, so connecting medical records to AI requires strong controls. OpenAI says the new healthcare environment includes controls such as role-based access, single sign-on and audit logs. Administrators can control which integrations are enabled, while existing user permissions determine what information an individual can access. OpenAI positions ChatGPT for Healthcare as an enterprise product designed for healthcare organisations and HIPAA-compliant workflows. The company has also made it clear that AI is intended to support healthcare professionals rather than replace them. Clinicians remain responsible for reviewing information and making care decisions. Why this matters The bigger development here is the growing connection between generative AI and the systems already used by hospitals. Instead of asking doctors to copy information from an EHR into a separate AI tool, integrations like this can bring AI closer to the existing clinical workflow. If implemented properly, this could reduce the amount of time clinicians spend searching through long patient records and preparing information before appointments. At the same time, healthcare organisations will need to carefully evaluate accuracy, privacy, security and human oversight before allowing AI to become part of important clinical workflows. For now, OpenAI's Epic integration is being introduced through supported healthcare deployments and pilot programmes.