ATA Action, the lobbying arm of the American Telemedicine Association, has filed comments on the FDA’s discussion paper covering generative AI in medical devices. The docket number is FDA-2026-N-7874 and the letter went to CDRH director Dr. Michelle Tarver, HIT Consultant reported.
The group’s sharpest argument is about the comparison point. Regulators traditionally score a device against what an attentive physician would do. In thousands of rural and low-income communities, ATA Action says, that physician is not the alternative. The alternative is a six-month wait, hours of driving, or no care. Joe Nye, who heads federal relations for the group, said the framework must stay clear and proportionate so tools reach underserved patients.
On structure, ATA Action backed the FDA’s two-axis matrix, which weighs how much clinical autonomy software exercises against how badly a wrong output could hurt someone. It asked the agency to regulate by function instead of by architecture, so a natural-language interface alone does not drag administrative tools into clearance.
One platform may do several jobs in a single visit, drafting a summary while flagging an abnormal lab. The group wants each function assessed on its own risk rather than the bundle forced through one fragmented review.
ATA Action also floated a voluntary federal-state sandbox, modeled on the FDA’s TEMPO pilot, so states writing their own clinical AI rules do not strand multi-state telehealth operators between conflicting regimes. It urged the agency to validate models across rural, low-literacy and non-English-speaking groups, to favor concise patient disclosures over long warnings, and to align with CMS on reimbursement.
