“For too long, our Department has been bogged down by bureaucracy and busy work; even the most productive public servants are mired in paperwork and process,” said Jim O’Neill, former deputy secretary of HHS.
HHS’s five pillars: an internal roadmap announced in December 2025
In December 2025 the Department of Health and Human Services (HHS) published a formal AI strategy organized around five explicit pillars: governance and risk management; infrastructure and platform design for user needs; developing the workforce and reducing burdens; fostering health research and reproducibility; and modernizing care and public health delivery. The department framed those pillars as the connective tissue for using AI across its departments and programs, with stated aims to expand access to shared tools and data and to accelerate manual approvals, claim adjudications, and grant review processes.
Federal guidance and the executive-order context since 2022
HHS’s strategy did not arise in isolation. The article notes a sequence of executive orders since generative AI entered the landscape in 2022, alongside guidance from the Office of Management and Budget, the National Institute of Standards and Technology, and the Cybersecurity and Infrastructure Security Agency. Agencies, the piece reports, have used those directives and guidance to establish agency-specific AI strategies intended to support mission goals while managing new technical and operational risks.
Centers for Medicare and Medicaid: AI as a force-multiplier against fraud, waste, and abuse
Reducing fraud, waste, and abuse (FWA) remains a top priority for federal health agencies. The Centers for Medicare and Medicaid (CMS), the article says, was recently tasked by an executive order with eliminating FWA and can apply AI and other technologies as part of a multi-pronged approach. The reporting highlights that modern fraud protection relies on deep data analysis powered by AI and machine learning to surface cross‑repository patterns and anomalies, and that those systems become more effective when agencies invest in data management and governance to ensure data are validated and usable.
FDA’s agentic AI rollout and the shadow of Elsa
The Food and Drug Administration has pushed forward with internal AI deployments. After launching a generative AI solution named Elsa in June 2025, the FDA has now deployed an agentic AI system available to all employees across the agency. Users of Elsa reported accuracy challenges—specifically that the model was prone to hallucination—and the article records those concerns. Despite that precedent, the piece quotes FDA officials as confident that the new agentic AI is equipped to enable complex workflows, assist with meeting management, support reviews and surveillance, and aid compliance and administrative functions. The agency’s agentic AI rollout, the reporting states, is part of a broader FDA AI strategy to streamline how the agency ensures the safety and efficacy of regulated products.
What this means for technologists and security teams, policymakers and regulators, and agency employees and applicants
- Technologists and security teams: The article underscores the importance of investing in data management and governance so AI-driven fraud detection can reference validated, usable data across repositories. At the same time, teams will be managing operational risks flagged by earlier systems—Elsa’s reported hallucinations being the example cited.
- Policymakers and regulators: Executive orders and interagency guidance are driving agency roadmaps; the piece implies continued reliance on OMB, NIST, and CISA guidance as agencies operationalize governance and risk-management pillars and balance detection with prevention for FWA.
- Agency employees and applicants: HHS frames AI as a tool to reduce paperwork and accelerate administrative processes—approvals, claim adjudications, and grant reviews—while CMS’s mandate to eliminate FWA positions AI deployments as a route to free up resources and improve service for legitimate applicants.
HHS has articulated a clear set of priorities and tools: five organizational pillars, reliance on cross‑agency guidance, and concrete deployments such as the FDA’s agentic AI. The essential measures the department will take next—expanding shared tools and turning guidance into operational practices that measurably reduce manual burdens and detect evolving fraud tactics—are stated goals. Whether those steps will fully overcome the accuracy and governance challenges foreshadowed by earlier systems like Elsa, and whether investments in data governance will translate into better outcomes for applicants and clinicians, remains the practical test of the strategy.
