HHS AI strategic plan implementation update: workforce training, equity, and safety priorities for 2026

HHS’s AI strategic plan was designed as a whole-of-department framework, and the implementation progress report reflects the uneven pace of execution across agencies with different cultures, budgets, and regulatory authorities. FDA’s AI implementation is the most mature, driven by a decade of building out the digital health regulatory function. CMS’s AI implementation is the most commercially consequential, given CMS’s role as the largest healthcare payer. NIH’s AI integration into biomedical research is the most scientifically ambitious. The Office of the National Coordinator for Health IT’s AI work is the most infrastructure-focused.

The equity review workstream is the one to watch most carefully. CMS administers programs that cover a majority of Americans, and AI tools embedded in those programs — prior authorization algorithms, quality-measure scoring, fraud detection — have the potential to amplify existing disparities or introduce new ones. HHS’s equity review framework for AI in CMS programs is designed to identify algorithmic disparities before they compound into systematic harm. The preliminary findings from the first cohort of AI tools reviewed are not yet public, but the methodology — demographic performance testing, disparity audits, feedback loops from affected communities — is being documented for later release.

The NIH biomedical research AI initiatives include a significant investment in building training datasets that are representative of the demographic diversity of the U.S. population. Historically, biomedical AI has been trained disproportionately on data from research institutions with patient populations that are wealthier, whiter, and healthier than the general population. Correcting that imbalance is a decade-long project, and HHS’s strategic investment is one of the few efforts working on it at scale.