State AI Laws Are Multiplying — But Federal Preemption May Swallow the Movement
TexTak holds an 18% probability on US Congress passing federal AI legislation — a number that reflects our read on a decade of congressional dysfunction on tech regulation, combined with the Trump administration's explicit deregulatory posture. Today's news complicates the picture in both directions simultaneously: state legislatures are moving faster and more substantively than we expected, which both increases the preemption pressure that could motivate federal action and demonstrates that the regulatory vacuum Congress refuses to fill is being filled anyway. Connecticut passed comprehensive AI legislation. Utah signed nine AI bills. The GUARD Act cleared the Senate Judiciary Committee unanimously. And the Trump administration is reportedly withholding BEAD broadband funds to freeze state AI laws. This is a pressure system building toward something — we're just not sure it resolves in the direction we've forecast.
Here's the honest tension in our 18% figure: it was built on the assumption that state-level momentum would be slow enough that Congress could defer indefinitely. That assumption is now under real pressure. Connecticut's SB 5 is comprehensive — frontier models, employment systems, data provenance, chatbot rules. Utah passed nine bills in a single session. Colorado is narrowing its 2024 law under legal challenge, but the legislative activity itself signals that state capitals are not waiting for federal clarity. When you have this much state-level movement, the compliance patchwork argument — historically the most powerful driver of federal preemption appetite — gets much stronger.
The Trump administration's broadband funding gambit is the most analytically interesting data point here. Virginia reportedly lost $800M in potential BEAD funds over AI regulation concerns. This is a real enforcement mechanism — not just rhetoric. But it also illustrates the structural problem with our federal legislation forecast: the administration's preferred tool is executive pressure, not legislation. If the White House can freeze state AI laws through funding conditions, the incentive to pass federal legislation drops further. Why go through the legislative process — with all its compromise and constraint — when executive leverage does the job? Our 18% already reflects this dynamic, but today's Virginia data suggests the executive route is more viable than we may have credited.
The GUARD Act is the most credible path to federal AI legislation, and the unanimous committee vote matters. Bipartisan unanimity on child safety is real — it's the same coalition that eventually passed COPPA, CIPA, and various platform liability reforms. But 'passed committee unanimously' has a poor conversion rate to 'signed into law' in the current Congress. The KOSA experience is instructive: overwhelming Senate support, stalled in the House, never signed. Our 18% accounts for this pattern.
What would move us above 30%: GUARD Act passing full Senate floor vote before August 2026. The child safety framing is the one lane where bipartisan momentum can survive the House, and a floor vote would signal the administration is willing to accept some legislative constraint on AI. What would keep us at 18% or push us lower: if the broadband funding pressure successfully freezes enough state AI laws that the compliance patchwork argument loses its urgency — removing the primary business community driver for federal preemption. We're watching the June 30 Colorado effective date closely. If that law is effectively neutered by federal pressure without federal replacement, the legislative pathway gets harder to see.