FDA and CMS Coordinate to Resolve Overlapping Jurisdiction on State AI Clinical Sandboxes

The Trump administration has expanded pathways for AI reimbursement through CMS while reducing certain FDA review burdens, yet states retain authority over unlicensed practice of medicine. This creates direct tension when AI systems operate without clinician supervision in sandbox environments, potentially classifying them as medical devices subject to federal premarket review. Officials have held closed-door discussions to map shared jurisdiction and prevent fragmented enforcement that could stall multi-state deployments.

Primary concerns center on AI performing autonomous clinical functions where state licensure laws intersect with FDA's 510(k) or De Novo pathways. Existing FDA guidance on AI/ML medical devices emphasizes lifecycle monitoring but does not explicitly address state-level experimentation programs, leaving gaps in liability and post-market surveillance. Analysis of similar past overlaps, such as telehealth cross-state licensing during COVID, shows that unresolved dual authority often results in slower adoption and higher compliance costs for developers.

Next steps likely include joint FDA-CMS guidance or model state legislation to create safe harbors for approved AI systems. Without such alignment, states with aggressive sandbox laws risk preemption challenges, while developers may limit testing to federally cleared use cases only. A unified framework would need to define thresholds for when AI crosses into regulated medical practice.

Evidence from regulatory filings and state health department announcements indicates preliminary talks have begun, though no formal interagency agreement has been released.