OpenAI said on September 1, 2026 that healthcare organizations can now pull authorized patient context from Epic into ChatGPT for Healthcare and query official public datasets from the same workspace. The company is pairing an electronic health record integration with a Healthcare Public Data plugin that reaches sources such as PubMed, DailyMed, and CMS Coverage.

The EHR path is meant to cut the tab-hopping that sits between a chart and a visit. OpenAI’s examples are the questions clinicians already ask: what changed since the last appointment, which labs matter today, whether medications or specialist notes moved, and which follow-ups are still open. ChatGPT is supposed to gather the authorized record, summarize the changes, and point back to the supporting chart.

The integration has two shapes. Teams can bring authorized Epic context into ChatGPT to review history and prepare for a visit. In supported deployments, ChatGPT can also sit inside the EHR layout so staff do not leave the chart.

Public data and a governed workspace

The Healthcare Public Data plugin is the other half of the announcement. OpenAI says it gives structured access to official healthcare datasets so a team can compare trial eligibility, check a drug label, or line up Medicare coverage language without searching each site separately. A research group could look up recruiting trials on ClinicalTrials.gov. A pharmacy team could confirm the latest DailyMed warnings. A population-health group could put research, open trials, and coverage rules in one source-backed view.

OpenAI says it works with hundreds of physicians across 60 countries, 49 languages, and 26 specialties, and that those reviewers have scored more than 700,000 model responses. For connected EHR context, physicians rated 4,363 answers across 27 clinical use cases, including pre-visit review, timelines, medication review, and handoff summaries. OpenAI reports that 99.1 percent of those responses were rated safe. In a separate two-round test on large U.S. healthcare datasets, more than 93 percent of answers for each of five connected sources were rated as having good or better accuracy.

ChatGPT for Healthcare sits behind enterprise controls: role-based access, single sign-on, and audit logs. With a Business Associate Agreement, OpenAI says customers can run ChatGPT Work, Codex, apps, and plugins in a HIPAA-compliant workspace. Individual U.S. clinicians on ChatGPT for Clinicians can install the public-data plugin. The Epic integration is not available on personal accounts.

“At AdventHealth, the value of AI starts with our people. By putting tools like ChatGPT Work, Codex, and connected business data in the hands of our team members, AI can reduce routine work and help practical innovations move more quickly into action. The goal is simple: give caregivers and teams more time for the human connection that enables our connected, whole-person care.” Robert Purinton, chief AI officer, AdventHealth

Decoded Take

Hospitals have heard the “summarize the chart” pitch before. The part that matters here is where the context lives: Epic as the record of record, official public sources as the citation layer, and a BAA as the permission slip. That is OpenAI trying to become the review pane on top of systems it does not own, rather than another inbox for discharge notes. The 99.1 percent safety rating is a vendor metric on vendor-chosen cases. Watch whether named health systems turn the Epic connector on for pre-visit work, whether those summaries stay cited to the chart, and whether compliance teams treat the workspace as in-scope for HIPAA audits. If the integration stays in a sidecar tab, it is a demo. If it sits in the EHR layout and changes how a clinic starts the day, it is infrastructure.