OpenAI has begun integrating ChatGPT Health with Epic’s electronic health record (EHR) system, a move that gives clinicians read-only access to patient data across the platform’s network of more than 325 million patient records. The integration, announced today, lets doctors and nurses pull up lab results, medications, appointment notes, and specialist documentation, then use AI to summarize and query that information directly within their existing workflow.
In some deployments, clinicians can access ChatGPT without leaving a patient chart, using it for pre-visit review, building clinical timelines, and identifying changes in a patient’s history. OpenAI said the integration is strictly read-only — the AI does not write anything back to the health record — and requires a Business Associate Agreement for organizations to use ChatGPT Work, Codex, apps, and connectors in compliant workflows.
Also read: Apple Says Returned MacBook Shows Ex-Employee Used Trade Secrets at OpenAI
How the Epic integration works in practice
The integration is designed to reduce the administrative burden that contributes to clinician burnout. Instead of toggling between multiple screens to piece together a patient’s history, clinicians can ask ChatGPT to summarize a chart or flag notable changes. For example, a physician preparing for a follow-up visit could prompt the tool to list all medication adjustments over the past six months or summarize the latest specialist’s notes.
OpenAI also introduced a Healthcare Public Data plugin that pulls from public sources including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. The plugin is meant to help healthcare workers synthesize data for trial eligibility criteria, medication identifiers, coverage policy versions, and provider records — tasks that typically require cross-referencing multiple government and clinical databases.
The rollout follows last month’s launch of ChatGPT for health to all U.S. consumers, which OpenAI said has driven roughly 300 million health-related queries per week. That consumer-facing expansion set the stage for deeper clinical integrations, but it also amplified scrutiny of the technology’s safety profile.
Safety concerns and legal challenges persist
OpenAI said it collected over 4,300 responses from physicians across 27 clinical use cases — including pre-visit review, clinical timelines, medication review, and handoff summaries — and found that 99.1% of responses were safe. That figure, while high, leaves room for error in a domain where even a single mistake can have serious consequences.
The company has repeatedly stated that AI is not suitable for diagnosis or treatment, and the Epic integration is positioned as a workflow tool rather than a clinical decision-maker. But legal challenges continue to mount. A few days before this rollout, a Florida-based pastor sued OpenAI, alleging that ChatGPT gave him a near-fatal recommendation. In May, family members of a user filed a separate lawsuit blaming ChatGPT for wrongful advice related to medication dosage.
These cases highlight the tension between the efficiency gains of AI-assisted chart review and the liability questions that arise when a model produces unsafe output. Even if 99.1% of responses are deemed safe by physician reviewers, the remaining fraction can be consequential in a healthcare setting.
What this means for clinicians and health systems
For health systems already using Epic, the integration offers a way to test AI capabilities without overhauling their existing infrastructure. The read-only design reduces the risk of accidental data modification, and the requirement for a Business Associate Agreement ensures that organizations using the tool are bound by HIPAA-related obligations.
But the rollout also raises practical questions. Clinicians will need training to interpret AI-generated summaries critically, and health systems must establish protocols for when and how to use the tool. The 99.1% safety figure, while reassuring, is not a guarantee — and the pending lawsuits serve as a reminder that legal exposure remains a real concern for both OpenAI and the institutions that deploy its technology.
OpenAI’s broader push into healthcare comes as the company expands its enterprise offerings across multiple sectors. The Epic integration is a significant step toward embedding AI into clinical workflows, but it is not a substitute for clinical judgment. As more health systems adopt the tool, the focus will likely shift to measuring real-world outcomes and establishing clear guardrails for AI use in patient care.
