Bipartisan AI Bills Are Multiplying — But Multiplication Is Not the Same as Passage

Let's give this news its full due before we explain why we're not moving. The American Leadership in AI Act is not performative. Lieu and Obernolte co-chair the House AI task force — this isn't a backbencher press release, it's the institutional leadership of House AI policy putting a comprehensive marker down. The 20+ proposals cover substantive ground: standards and evaluation, R&D infrastructure, federal AI adoption frameworks, and worker support. The deepfake provisions specifically have bipartisan emotional resonance that could survive conference. This is more organized and more substantive than anything we've seen from Congress on AI in the prior two years.

But our 18% doesn't reflect skepticism about whether serious people are drafting serious bills. It reflects the structural gap between bill introduction and presidential signature, which in the current environment requires navigating: (1) Senate action with 60-vote threshold on anything controversial, (2) a White House with a stated deregulatory philosophy that views federal AI mandates as industrial policy interference, and (3) a legislative calendar that shrinks materially as 2026 midterm positioning consumes bandwidth. The American Leadership in AI Act is explicitly a pro-innovation, pro-standards frame — it's not a regulatory crackdown — which makes it more compatible with the administration's instincts than a pure safety bill would be. But 'compatible' and 'prioritized for floor time' are very different things. Congress has not passed comprehensive technology regulation in a decade, and that failure mode has survived multiple moments of bipartisan consensus formation.

The strongest counterargument to our 18% is that the AI bill isn't framed as regulation — it's framed as competitiveness infrastructure. Funding R&D, setting standards, modernizing federal adoption. That framing has historically gotten further in Republican Congresses than safety-and-liability frames. The national security angle on the deepfake provisions is real. If proponents can separate the workforce and standards provisions from anything that looks like industry mandate, pieces of this bill could move as riders on appropriations or defense authorization vehicles. We genuinely haven't fully priced the possibility that a narrow slice of this legislation — not the full bill — gets attached to something that does pass. But our forecast target is 'Congress passes federal AI legislation,' which we've interpreted as a meaningful standalone act or comprehensive rider, not a single deepfake provision buried in NDAA.

What would move us to 30%+: The Senate companion bill drops within 60 days with Republican primary sponsors, AND at least one of the 20+ provisions gets scheduled for committee markup in the Senate. What would confirm we're right to hold at 18%: The bill sits in committee through the August recess without a Senate counterpart, which historically signals a messaging exercise rather than a legislative effort. We're watching the Senate response window, not the House activity we've already seen.