Federal AI Legislation Is Functionally Dead for 2026 — But the State Pressure Building Beneath It Is the Real Story
textak holds US federal AI legislation at 3%, and today's news doesn't move that number — it explains it. The Congressional AI agenda for Q3 2026 has defaulted to reactive fragments: chatbot safety for minors, algorithmic pricing, data center power demands. Connecticut's CART Act takes effect October 1 without any federal framework in place. Congress is responding to state momentum it cannot preempt because it cannot agree on whether it wants to. This is not a legislative body working toward a comprehensive AI bill. It's a body managing the political fallout of not having one.
Our 3% reflects 77 days remaining, zero Republican co-sponsors on the Senate AI Risk Management and Security Act, maximum veto certainty from the Trump administration, and a preemption debate so internally fractured that a 10-year moratorium on state regulation — which would at least have been coherent — couldn't survive its own caucus in 2025. Today's ML Strategies reporting confirms the deadlock is structural, not tactical. It's not that negotiators are close and need more time. It's that the Republican caucus cannot agree on the basic architecture of federal involvement, which means even a bill that passes committee faces a cloture problem in the Senate and a veto problem at the White House.
The honest pressure test here is whether we're underweighting the state-pressure forcing function. Connecticut's CART Act is now live enforcement with real compliance obligations — automated employment decisions, AI companion chatbots, frontier model developers, attorney general enforcement authority. That's not a pilot. Twelve states with AI personhood exclusion laws, state-level AI liability frameworks proliferating, and now staggered compliance deadlines through 2028 in one state alone. The business community that historically pushes for federal preemption to replace state patchwork has a stronger lobby case than ever. Why doesn't that move us above 3%?
Because the mechanism is wrong for the timeline. Business lobby pressure for federal preemption is real but it operates on a multi-year horizon — the kind of pressure that produces legislation in 2028 or 2029 after the patchwork becomes operationally untenable. For 77 days, you'd need bicameral passage and a presidential signature. That requires: a bill with bipartisan support (doesn't exist), floor scheduling in both chambers (not happening), and a White House that reverses its executive-order-as-substitute approach (zero signal of this). State momentum is building the case for eventual federal legislation. It is not compressing the timeline for this year's resolution.
The part of our model that genuinely concerns us is the 'executive order as substitute' pathway. If Trump issues a sweeping AI executive order in November or December that preempts state action through administrative rather than legislative mechanism, does that satisfy 'Congress passes federal AI legislation'? No — by definition. But it would be a substantive policy event that our forecast would miss entirely. We're watching for executive branch action that makes our 3% technically correct but analytically incomplete.