GAAIA and Ten House Bills: Congress Gets Specific About AI

For a couple of years, the story on federal AI legislation was that there wasn't much of one. That changed in June 2026, and it changed on two fronts at once. A sweeping 269-page framework arrived as a discussion draft, and, separately, a committee quietly advanced ten smaller, targeted bills. Reading them together is the useful exercise, because they represent two very different theories of how Washington should legislate AI — the comprehensive framework and the incremental fix. Here is a plain, attributed walk through both.

Two tracks at once

The first track is the Great American AI Act, or GAAIA, a bipartisan discussion draft aiming to establish an overarching federal regime for AI. The second track is a batch of ten narrower bills that the House Science Committee marked up and reported favorably on June 25. The distinction matters: a discussion draft is an invitation for feedback, not a bill headed for a floor vote, while a favorably reported bill has cleared a real procedural hurdle. Conflating the two — treating the sweeping draft as if it were about to become law — is the most common mistake in coverage of this moment.

GAAIA: the big framework draft

The Great American AI Act was released by Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA), with several colleagues joining, as a 269-page draft. According to the sponsors' own announcement and analyses such as DLA Piper's, the draft is organized into four titles: frontier AI governance, workforce, cybersecurity, and research, development, and international cooperation. It is being circulated to gather feedback before any formal introduction.

Two features draw the most attention. First, the draft defines who is covered by size: a "frontier developer" is broadly an entity that has trained a frontier model and had more than $50 million in prior-year gross revenue, while a "large frontier developer" is the same but above $500 million. That thresholding means the heaviest obligations would fall on the largest developers, not on small teams or startups. Second, the draft proposes preempting state laws that specifically regulate the development of AI models, with a three-year sunset, while leaving state laws about the use or deployment of AI in place, as Roll Call reported. The preemption provision is the most contested part; supporters frame it as avoiding a patchwork, and critics worry it narrows state authority. We are flagging the debate, not taking a side in it.

The ten bills that actually moved

While GAAIA gathers comment, the House Science Committee advanced ten smaller bills in a single markup. Per a roundup from Mintz, most passed unanimously, and several are aimed squarely at the plumbing of federal AI — standards, data, workforce, and research access rather than sweeping mandates.

The ten bills reported June 25, 2026

Bill	What it would do	Vote
H.R. 2385 — CREATE AI Act	Establish the National AI Research Resource (NAIRR) at NSF to widen access to compute and datasets	29-0
H.R. 9363 — AI Security and Innovation Act	Strengthen the security of federal AI systems	29-0
H.R. 9341 — AI-Ready Federal Data Guidelines Act	Direct NIST to help agencies prepare datasets for AI	29-0
H.R. 6461 — READ AI Models Act	Direct NIST to build documentation and transparency templates	35-0
H.R. 9333 — AI Flaw Reporting and Security Enhancement Act	Create a voluntary NIST program to track AI vulnerabilities	35-0
H.R. 8893 — Protecting Consumers from Deceptive AI Act	Support research on detecting and authenticating AI-generated content	35-0
H.R. 9334 — Workforce for AI Trust Act	Expand NIST AI-workforce activities and skills frameworks	35-0
H.R. 5351 — NSF AI Education Act	Expand NSF AI education, scholarships, and fellowships	33-0
H.R. 5584 — LIFT AI Act	Authorize NSF support for K-12 AI literacy	34-1
H.R. 9372 — Data Infrastructure Energy Measurement Act	Direct NIST and DOE to improve data-center energy and water metrics	34-1

The headline item is the CREATE AI Act. H.R. 2385 would formally establish the National Artificial Intelligence Research Resource — NAIRR — housed at the National Science Foundation to expand access to computing power and datasets for AI research. As Congress.gov records, the bill has bipartisan roots, and the committee reported it 29-0. NAIRR's purpose is to broaden who can do serious AI research beyond the handful of institutions that can afford large compute, which is a supply-of-talent story as much as a supply-of-hardware one.

Why the narrow bills may matter more

It is tempting to focus on the 269-page draft, but the ten bills may be the more consequential news, precisely because they are modest. Narrow, bipartisan bills that direct NIST to write guidelines or NSF to fund research tend to move; comprehensive frameworks that reallocate authority and preempt state law tend to stall, get rewritten, or wait. The committee votes tell the story — near-unanimous margins signal these are not the fights, whereas GAAIA's preemption clause is exactly the kind of provision that generates a fight.

What it means in practice

For most builders and small firms, the immediate reading is reassuring: GAAIA's heaviest obligations are scoped to developers above $50 million and $500 million in revenue, so a startup training on top of existing models is not the target. The bills that would touch a wider audience soonest are the standards-and-data ones — if NIST publishes AI documentation templates or data-readiness guidelines, those become de facto reference points well beyond the agencies they are written for. If your work intersects government, our federal AI readiness guide and AI for government overview are useful for tracking how this kind of standards work lands, and our free course on federal AI covers the landscape in more depth.

The larger point is that "Congress and AI" is no longer a single story. There is a maximalist track (GAAIA) that is genuinely uncertain, and an incrementalist track (the ten bills, plus NAIRR) that is quietly advancing. The useful habit is to hold both in view without collapsing them — to notice that the sweeping draft is a signal of intent while the small bills are the ones to actually watch move. Track the votes, not the page counts.

Sources: Rep. Obernolte — GAAIA discussion draft; DLA Piper; Roll Call; Mintz — AI Washington Report; Congress.gov (H.R. 2385). Framing by Precision AI Academy; positions attributed to sponsors and cited analysts.

Common questions

Is GAAIA now law? No. It is a discussion draft released for feedback, not a bill that has been introduced or voted on. Treat it as a proposal and a signal of direction.

Does GAAIA regulate every AI company? Its heaviest obligations are scoped to "frontier developers" above $50 million in prior-year revenue and "large frontier developers" above $500 million, so most small teams fall outside the core requirements as drafted.

What is NAIRR? The National Artificial Intelligence Research Resource, which the CREATE AI Act would establish at the National Science Foundation to give researchers wider access to computing power and datasets.

Which of these is most likely to become law? No one can say for certain, but the ten committee-reported bills — especially the near-unanimous ones — have cleared a real hurdle that the GAAIA draft has not.