Great American AI Act Draft: Federal Frontier Framework Meets Three-Year State Preemption Fight
Obernolte–Trahan 269-page bipartisan discussion draft proposes binding obligations for large frontier developers (~$500M+ revenue in analyses), workforce and cyber titles, and multi-year preemption of certain state AI development laws. Parallel—and partially conflicting—path vs EO 14409’s voluntary track.
If advanced, GAAIA would be the first comprehensive federal AI statute of its kind—directly regulating frontier labs while reordering federal-state power over AI rules. Released days after EO 14409’s voluntary frontier-access track, it shows Congress and the White House pursuing parallel, sometimes conflicting, models of national AI governance at the exact moment labs file for IPOs.
House members released a bipartisan discussion draft of the Great American Artificial Intelligence Act of 2026. Led by Reps. Jay Obernolte and Lori Trahan, the ~269-page proposal aims at a single national framework for frontier AI governance, worker protections, cybersecurity, and R&D—while preempting a patchwork of state rules on AI development for a multi-year period. It is not yet formally introduced; sponsors are soliciting feedback from industry, states, workers, and civil society.
What’s in the draft
Primary materials from Obernolte’s office and contemporaneous legal analyses describe four major titles:
- Frontier AI Governance — Binding development/transparency-style obligations aimed at “large frontier developers” (analyses cite a threshold around $500M+ annual revenue for firms that have trained a frontier model—capturing labs such as OpenAI, Anthropic, Google, Meta, and xAI if enacted as drafted).
- Workforce — Provisions addressing labor-market impacts of AI adoption.
- Cybersecurity — Federal posture on AI-related cyber risks.
- Research, Development, and International Cooperation — Capacity-building and coordination measures.
A central political feature is time-limited federal preemption of certain state laws related to AI development (reported as a three-year preemption window), framed by supporters as avoiding fifty different state regimes. The draft draws conceptual lineage from state frontier-model transparency laws in California, New York, and Illinois.
Cosponsors joining the release: Scott Franklin (R-FL), Suhas Subramanyam (D-VA), Erin Houchin (R-IN), and Scott Peters (D-CA). The House Democratic Commission on AI publicly opposed the draft hours after release, underscoring unsettled partisan fault lines even on a bipartisan draft.
Sponsors published a Bloomberg Law op-ed the same day calling the text “the start of a serious national conversation,” not a final product.
Limits
- Discussion draft only—not introduced, marked up, or enacted.
- $500M frontier-developer threshold and three-year preemption window are analyses of draft text; final bill text can change.
- Same-day Democratic Commission opposition shows bipartisan sponsorship ≠ floor consensus.
Sources
- Rep. Jay Obernolte press release: “Obernolte, Trahan release a discussion draft of the Great American AI Act” (obernolte.house.gov, June 4, 2026). Primary announcement with cosponsor quotes and PDF links.
- Discussion draft PDF and section-by-section summary linked from the release.
- Roll Call, DLA Piper, TechPolicy.Press contemporaneous analysis (June 4–5, 2026) on preemption, title structure, and $500M frontier-developer framing.