U.S. pushes federal AI bill
Congressional Republicans introduced the SMART AI Act to target large model developers with new safety-disclosure rules and federal preemption of state laws — the US’s first big federal swing at AI regulation. The move lands into a fragmented global debate — the White House is reportedly pressuring Republicans to curb state-led rules while the UK, UAE, Kenya and the UN are all pursuing different frameworks, underscoring how messy international harmonization will be ( )
Sen. Marsha Blackburn is the architect of the proposal, which she first filed as a draft framework in mid‑2025 and bills the SMART AI Act as “Supporting American Innovation and Responsible Technology.” (webpronews.com) The draft establishes compute‑based tiers—using FLOPs thresholds to classify “frontier” models—and assigns the heaviest compliance obligations to the highest‑compute developers, including mandatory safety evaluations, red‑team testing and public disclosures to a proposed AI Safety Board within the Department of Commerce. (webpronews.com) Buried language in related bills and analyses would preempt state and local AI rules for an extended period (analysts flag a ten‑year federal preemption in some versions), even though Congress previously rejected a moratorium on state AI laws by stripping that provision from a major package after a 99–1 vote in the Senate last year. (holonlaw.com) The White House has moved to codify a single national AI blueprint and publicly urged Congress to block state laws, releasing its legislative framework on March 20, 2026, but that push has provoked organized resistance from over 50 Republican state lawmakers and complaints from some House Republicans who say the administration’s pressure chills state‑level action. (whitehouse.gov) That U.S. effort arrives amid sharply different approaches overseas: the UK is advancing an Artificial Intelligence (Regulation) Bill and debating a principles‑led statutory model, while the UAE has issued a national AI charter and layered rules across federal and free‑zone authorities. (bills.parliament.uk) African and multilateral tracks are separate still—Kenya rolled out a 2025–2030 National AI Strategy rather than a single AI statute, and the United Nations has launched an Independent International Scientific Panel and a Global Dialogue on AI Governance that will convene international technical and policy coordination starting with events planned for 2026. (taxnews.ey.com)
