Federal health agencies are moving fast on healthcare AI. At a CTA-hosted event in Washington, D.C., CMS Deputy Administrator Stephanie Carlton laid out the agency's four-pillar AI strategy: building public trust, expanding data sharing and interoperability, creating clearer market access pathways, and developing reimbursement frameworks for AI-enabled technologies. A key centerpiece is the ACCESS model — a 10-year, value-based chronic care program using AI to manage conditions like diabetes, hypertension, and depression across 150+ participating organizations. Carlton envisions the model demonstrating a "massively deflationary impact" on healthcare costs by 2028.

Meanwhile, the FDA is rethinking how it evaluates AI medical devices. The agency released a discussion paper in August proposing a "competency-based" framework for generative AI tools, with public comment open until October 19. FDA officials acknowledged the current review framework is a "square peg and round hole" for generative AI.

Not everyone is on board with moving quickly. Physicians debated the risks of fully autonomous AI, with AMA leaders stressing that clinical judgment — like noticing a patient's tremor during a routine refill visit — can't always be automated.

Key Takeaways:

• CMS is exploring paying tech companies directly for AI-driven care, a major shift from paying clinicians and facilities
• The FDA's existing review framework is considered ill-suited for generative AI and is under revision
• Clinicians warn that "human-in-the-loop" requirements must be evidence-based — not assumed — to avoid both over-reliance and unnecessary delays

Why it matters: The decisions being made now about AI regulation, reimbursement, and oversight will shape how — and how safely — AI enters everyday clinical care for millions of patients.