HomeDealsNegotiationsPolicyPipelineMoneyPeopleDataThe WeekPharmTech 100Deal TrackerResearchCompany Lookup

FDA’s AI Framework Debate Is Really About How Much Change Regulators Can Tolerate

KFF’s latest AI-in-health-care episode zeroed in on a basic problem: the FDA’s 1976 device framework was built for static products, not systems that learn.

By RxInsider Editorial · Sep 17, 2026 · 342 words · via KFF Health News
FDA’s AI Framework Debate Is Really About How Much Change Regulators Can Tolerate

Image: KFF Health News

What happened

On September 15, 2026, KFF Health News published an episode of The Business of Health with Chip Kahn focused on how the FDA regulates artificial intelligence in health care. The discussion featured Brian Miller, MD, an associate professor at the Johns Hopkins University School of Medicine and a visiting fellow at the Hoover Institution, who has also worked at the Centers for Medicare & Medicaid Services, the Federal Trade Commission, the Federal Communications Commission, and the FDA. KFF framed the core issue as what happens when the FDA applies its 1976 medical device framework to AI, “a technology that learns and changes.” Recorded shortly before the FDA released a discussion paper outlining possible regulatory approaches and inviting stakeholder feedback, the episode closed with Kahn weighing in on that paper.

Why it matters

The immediate takeaway is less about one policy paper than about a structural regulatory mismatch. Miller’s central argument, as described in the source, challenges the assumption that today’s manual system is inherently safer or more consistent. What keeps him up at night, the source says, is not the speed of AI but the risk that “fear-driven regulation” could cost the system the opportunities AI offers. That framing matters because it suggests the next phase of FDA oversight may turn on how much product change regulators are willing to tolerate after launch, not simply whether AI belongs inside the existing device rubric.

For industry, one likely read is that developers will be watching the discussion paper for signals on how the FDA distinguishes between acceptable iteration and changes that trigger heavier review. For payers and health system buyers, the issue extends beyond approval mechanics. If regulatory policy slows deployment of tools that promise more consistency than current manual workflows, adoption curves could flatten even where operational demand is strong. Investors should also pay attention to the overlap between FDA policy and Medicare payment thinking, given Miller’s background across both regulatory and payment institutions. Detailed drug monographs are at ClinicalRx.ai. For adjacent drug and policy developments, see RxNews.ai.

RxInsider combines reported facts with industry analysis and informed inference. Forward-looking reads, market commentary, and interpretive framing reflect analysis of available reporting and known facts, not confirmed outcomes.

Tags
theweekformat:briefingsynthesispolicyresearch
The Insider - Weekly pharma intelligence
Deals, negotiations, and policy analysis. Delivered when it matters.
No sponsored content. No noise. Unsubscribe anytime.
More from The Week
All The Week →
The WeekFiercePharma ↗
After a 2024 rebuff, AstraZeneca and Daiichi Sankyo's star antibody drug conjugate Enhertu has been deemed cos…
Sep 17, 2026
The WeekFierceBiotech ↗
A phase 2b study of Longeveron's stem cell therapy in a pediatric heart disease has missed its primary endpoin…
Sep 17, 2026
STAT+: A biotech hedge fund births a new immunology company
The WeekSTAT News ↗
Why a biotech hedge fund decided to build from the ground up a new company to develop a better, long-acting an…
Sep 17, 2026