OpenAI is making its most aggressive move yet into clinical workflows, announcing an integration that puts ChatGPT Health inside Epic’s electronic health record system — the software that holds data for over 325 million patients. Clinicians will be able to pull appointment notes, lab results, medications, and specialist documentation into ChatGPT for summarization, pre-visit review, and building clinical timelines without ever leaving a patient’s chart. The access is strictly read-only; the AI can’t write anything back into the record.
The company is also launching a Healthcare Public Data plug-in that pulls from ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. For clinicians, that means synthesizing trial eligibility criteria, medication identifiers, and coverage policy versions on demand. Organizations with a Business Associate Agreement can now use ChatGPT Work, Codex, apps, and connectors for HIPAA-compliant workflows. The timing is notable: last month OpenAI made ChatGPT for health free to all U.S. consumers and disclosed that people fire off 300 million health-related queries weekly.
OpenAI says it collected responses from more than 4,300 physicians across 27 clinical use cases and found 99.1% of AI outputs were safe. That’s a strong number, but the denominator matters more than the percentage. With 300 million weekly health queries, even a 0.9% unsafe response rate translates to millions of potentially dangerous answers floating around. OpenAI has been careful to maintain that the tool isn’t for diagnosis or treatment — a disclaimer that’s getting harder to take at face value as the product moves deeper into clinical settings.
This rollout lands days after a Florida pastor sued OpenAI, alleging ChatGPT gave him a near-fatal recommendation, and months after a family blamed the chatbot for wrongful dosage advice that they say contributed to a death. The Epic integration is technically impressive and genuinely useful for reducing chart-review drudgery. But shipping it while those lawsuits are still active sends a clear signal about where OpenAI’s priorities sit: enterprise adoption first, liability questions later.
