Great American AI Act Discussion Draft Proposes 3-Year Preemption of State AI Laws; White House Bundles It With KOSA Package

A bipartisan 269-page discussion draft released June 4, 2026 by Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) — the Great American AI Act — would impose federal AI transparency and safety obligations on large developers while preempting state AI laws for three years. Simultaneously, the White House is negotiating directly with Senate leadership to bundle three-year AI preemption with KOSA, the NO FAKES Act, and a federal age-verification mandate. The administration’s December 2025 AI executive order had already created an AI Litigation Task Force to challenge conflicting state laws. More than 600 state AI bills are active, with California, Colorado, and Texas holding the most advanced frameworks now at risk; skeptics in both parties question whether sweeping preemption is achievable before the November 2026 midterms.

This is the legislative companion to EO 14365’s litigation-driven preemption strategy (2025-12-11–eo-14365-ai-preemption-doj-litigation-task-force-against-state-ai-laws) and the AI Action Plan that preceded it (2025-07-23–ai-action-plan-released): the executive branch sues to strike state laws while Congress is steered to freeze them outright. The preemption move follows the historical template of the OCC’s preemption of state predatory-lending laws (2004-01-13–office-comptroller-currency-preempts-state-predatory-lending-laws) and ERISA’s benefits preemption (1974-09-02–erisa-enacted-corporate-benefits-preemption) — using a federal floor to wipe out stronger state protections. The bundling with KOSA is the leverage; the first state laws the Litigation Task Force challenges will operationally define what the administration treats as “obstruction of national AI policy.”