Source: Atlantic Council

Federal agencies reported rapid expansion of AI adoption, with 3,611 individually reported AI use cases as of April 2026, including 445 classified as high‑impact across forty‑one agency submissions; procurement and integration have been facilitated by the General Services Administration, enabling access to frontier models from providers such as Anthropic, Google, OpenAI, and xAI. Policy guidance has evolved through OMB memoranda M‑25‑21 and M‑25‑22 (April 2025) and M‑26‑04 (December 2025), which introduced structured approaches to federal AI adoption, designated chief AI officers, and added procurement and transparency requirements for generative AI systems; the Department of Defense launched an AI Acceleration Strategy in January 2026 to integrate advanced AI capabilities into military operations.

Agencies confront operational and governance challenges including legacy cybersecurity and privacy frameworks that can impede interagency data sharing, NIST’s Generative AI Profile (AI 600‑1) identified risks such as hallucinated outputs and information‑integrity concerns, and there are gaps in standards for model provenance, data interoperability, and classification controls. Federal policy remains a patchwork of sectoral rules, voluntary guidance, and procurement policy while states adopt their own AI laws, producing fragmentation and tension with federal efforts including a December 2025 executive order that contemplates preemption. The report identifies findings and recommendations: creating a coherent federal regulatory framework with mandatory and voluntary standards for risk tiering, procurement, data governance, and transparency; reforming procurement to favor outcomes and continuous evaluation; investing in workforce training and the US Tech Force (~1,000 technical experts for limited terms); and aligning AI adoption with national defense and international industrial policy, including export controls and the 2025 AI Action Plan to scale U.S. technology abroad.