OpenAI is connecting ChatGPT Health directly to Epic's electronic health record system, which manages data for over 325 million patients. This lets doctors pull patient information—appointment notes, lab results, medications, specialist reports—and have ChatGPT summarize and organize it without leaving their EHR interface. The system works read-only, so the AI can't write anything back to patient records. In some setups, clinicians can do pre-visit reviews and build clinical timelines straight from the patient chart without switching between systems. The company also rolled out a Healthcare Public Data plug-in that taps into ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. This lets healthcare workers quickly cross-reference trial eligibility, medication details, coverage policies, and provider information. OpenAI tested the integration across 27 clinical use cases with over 4,300 physician responses and reported a 99.1% safety rate. That sounds solid on paper, but the company acknowledges that even a small percentage of bad answers can cause real harm—which matters when you're dealing with health decisions. The timing here is worth watching. OpenAI launched ChatGPT for health to U.S. consumers last month and says people ask it 300 million health questions weekly. But the company faces ongoing legal pressure: a Florida pastor sued over a near-fatal recommendation he received, and in May, a family sued claiming ChatGPT gave dangerous dosage advice. OpenAI keeps saying the tool isn't meant for diagnosis or treatment, yet it's being embedded deeper into actual clinical workflows where those lines get blurry fast.