OpenAI Makes Epic Integration Available for ChatGPT for Healthcare

ChatGPT Epic integration has been available since September 1, 2026, for eligible organizations using ChatGPT for Healthcare and select HIPAA-enabled ChatGPT Enterprise workspaces. It allows an Epic electronic health record (EHR) environment to be connected to the chatbot so clinicians can work with data they are already authorized to access.

However, this is not a feature for the consumer ChatGPT Health service, nor is it available to individual ChatGPT for Clinicians accounts. Deployment requires an approved organizational workspace, administrator configuration of the Epic application, an individual user login to Epic, and the relevant contractual and HIPAA settings.

ChatGPT Epic integration is read-only

The connection is designed as read-only access. It cannot edit health records, create orders, or change existing access rights to patient charts. ChatGPT therefore receives only the scope of data that a specific user can also view directly in the Epic system.

In practice, the tool is intended to help process information from the permitted portion of documentation. Intended uses include summarizing notes, medications, laboratory results, and changes since the patient’s previous visit. It is not a standalone health-record system or a mechanism that can write back to the EHR.

OpenAI also announced the Healthcare Public Data plugin. It is used to search public healthcare sources but does not access patient charts. It is therefore a separate feature from the Epic integration.

Access to protected data has organizational restrictions

Deploying generative AI with health records brings the tool directly into workflows involving protected health information. An important restriction is the preservation of existing permissions and the read-only mode, which rules out unauthorized changes to the EHR through the integration.

That does not eliminate the need for clinical review of the output. Even a system used for summarization can overlook important data or misinterpret information in a record. The final clinical decision remains with the healthcare professional; relevant information must be verified against the source documentation.

OpenAI reports its own evaluation, according to which physicians rated 99.1% of 4,363 responses across 27 clinical scenarios as safe. However, the company did not provide independent verification of the evaluation’s methodology or results. It is also not independently confirmed by the available information how many healthcare organizations have already deployed the connection or to what extent clinicians use it in routine practice.

What will be important to monitor

Further developments will show which hospitals and healthcare systems actually launch the integration and under what conditions it will be available outside the United States. Independent evaluations of accuracy, errors, effects on work time, and patient safety will also be relevant.

For individual deployments, specific rules for access auditing, PHI protection, and the scope of EHR sources or fields that an organization makes available will also be important. The feature’s availability will therefore not be the same for every user or every Epic system.

Verified and updated: 09/01/2026 22:30