Video Preview
🎧 Podcast episode for paid subscribers only. Also available on Apple Podcasts and Spotify.
To listen to paid episodes in Apple or Spotify, link your Substack subscription via the show settings on those platforms (instructions inside the Substack app under Subscriptions → Podcast).
Table of Contents
What actually shipped on September 1
The plumbing is boring and that is the whole story
Epic’s very loud silence
Three AIs, one chart, zero referees
The liability question nobody has priced
Governance, HIPAA, and the opt out that does not exist
Trial matching and the pharma subplot
What to actually do with this
Abstract
OpenAI announced on September 1, 2026 that ChatGPT for Healthcare now connects to Epic environments, read only, with two modes: pull chart data into ChatGPT, or embed ChatGPT inside supported Epic workflows
A companion Healthcare Public Data plugin wires in official sources including ClinicalTrials.gov, PubMed, RxNorm, DailyMed, CMS Coverage, and other CMS datasets
This is not a partnership. It rides Epic’s free open FHIR APIs, the same on-ramp used by thousands of other apps. Epic did not co-announce; its public comments frame the integration as one of those thousands of apps that health systems can choose to enable.
UCSF Health is piloting. AdventHealth leaders have commented on the broader OpenAI healthcare workspace. OpenAI is publishing safety numbers: 99.1 percent of responses rated safe across 4,363 physician ratings, and connector accuracy from 93.2 to 98.6 percent depending on the public data source.
The interesting questions are downstream: liability when a summary omits the one med that mattered, patient opt out (there isn’t one), FDA clinical decision support boundaries, and the fact that most large health systems already pay two other vendors to do exactly this job
