The Centers for Medicare and Medicaid Services wants to build a more consistent way to pay for clinical AI and software, proposing changes for 2027 that factor in a technology’s impact on patient outcomes.
In proposed rules for hospital outpatient payments and physician fees, CMS signaled readiness to move beyond the old model of treating algorithms like physical items. The agency has long struggled to value software-based tools — an algorithm that predicts cardiac risk from a CT scan or an AI map of prostate cancer spread doesn’t fit neatly into payment codes designed for cotton swabs and CT scanner wear.
For 2027, CMS proposes practical labeling and payment changes for several clinical AI services as an interim step. The broader goal is a standardized structure that accounts for whether these tools actually improve patient health, not just whether they were used.
The move follows years of debate over who should pay for clinical AI and how. Only three AI devices have received permanent CPT codes so far, leaving most hospitals and clinics navigating a patchwork of temporary billing arrangements. Industry groups including Epic, Oracle, Abridge, Aidoc and Tempus have submitted proposals to HHS on driving clinical AI adoption.
The proposed rules represent CMS’s most concrete signal yet that it intends to solve the AI payment puzzle — a shift that could determine whether AI tools become routine in American medicine or remain financially out of reach for most hospitals.
