Clinical AI on Trial: Shadow Tools and Governance Gaps | Newsday with Yaw Fellin and Matt Troup

July 20, 2026: A new Nature of Medicine study claims general-purpose LLMs outperform specialized clinical AI, but is that the right benchmark to begin with? Bill and Drex are joined by Yaw Fellin, General Manager of UpToDate at Wolters Kluwer, and Matt Troup, Clinical Strategy Principal at Abridge, to break down what those benchmarks actually measure, why shadow AI is already spreading through health systems, and what "clinical grade" really means. Plus: the first-ever look at UpToDate's new drug dosing feature in Expert AI, and why automaticity bias may be the next liability crisis hiding in plain sight.Key Points:03:18 Nature Study Breakdown10:25 Shadow AI Reality Check18:37 Transparency and Provenance23:40 Human in the Loop WrapKeep up to date on the latest in health IT:https://thisweekhealth.com/news/X: This Week HealthLinkedIn: This Week HealthDonate: Alex’s Lemonade Stand: Foundation for Childhood Cancer