A bipartisan group of House lawmakers unveiled a sweeping 269-page discussion draft on Thursday that would establish the first comprehensive federal framework for governing artificial intelligence in the United States — and in doing so, would temporarily override a growing patchwork of state AI laws.
Dubbed The Great American Artificial Intelligence Act, the bill was introduced by Reps. Jay Obernolte (R-Calif.) and Lori Trahan (D-Mass.), joined by four other co-sponsors. It comes just days after President Trump signed an executive order establishing voluntary federal reviews of frontier AI models, and as AI companies like Anthropic, OpenAI, Google DeepMind, and xAI face mounting scrutiny over safety risks.
The timing is no accident. AI development has moved so rapidly that dozens of states have enacted their own patchwork of regulations — and both industry players and some federal lawmakers argue that a confusing maze of 50 different rulebooks could stifle innovation and create compliance chaos.
What the Bill Would Actually Do
At its core, the Great American AI Act creates a new accountability structure for large frontier AI developers — those with more than $500 million in annual gross revenue. Companies in this category, which would include Anthropic, OpenAI, Google DeepMind, and xAI, would be required to:
- Publish public frontier AI frameworks that outline whether their models could pose “catastrophic risks,” defined as a foreseeable risk of death or injury to more than 50 people, or more than $1 billion in property damage.
- Report critical safety incidents to federal and state regulators within 15 days of discovery, and imminent catastrophic risks within 24 hours.
- Submit to independent audits by licensed Independent Verification Organizations every six months.
- Protect internal model weights through documented cybersecurity procedures.
- Face civil penalties of up to $1 million per day for violations.
The bill would formally establish the Center for AI Standards and Innovation (CAISI) in statute — previously known as the AI Safety Institute under the Biden administration — under the National Institute of Standards and Technology. It would be authorized $100 million per year for fiscal years 2027 through 2029 and empowered to evaluate both domestic and foreign AI systems for security vulnerabilities.
Notably, the bill would also extend the Cybersecurity Information Sharing Act of 2015 through 2035, allowing companies to share threat information without antitrust liability — a provision with implications well beyond AI.
The State Law Preemption Fight
Perhaps the most contentious element of the bill is its three-year preemption of state laws “specifically regulating the development” of AI models. The provision would sunset automatically after three years, but critics say the damage could be done long before then.
The bill’s own FAQ document singles out California laws as examples of what would be preempted: AB 2013, which requires model developers to publish summaries of their training data, and portions of SB 942 related to content watermarking. It also says frontier safety laws in California, New York, and Illinois would be “federalized” under the new framework.
What would NOT be preempted is equally important: state laws governing how AI is *used* rather than *developed* would remain fully intact. This means civil rights protections, labor laws, consumer privacy regulations, and laws against child sexual abuse material would still apply at the state level. State attorneys general would also retain enforcement powers over the federal safety requirements.
But critics are not mollified. Brad Carson, president of Americans for Responsible Innovation, called the preemption provision a “generational mistake.”
“This bill takes the current floor on state AI legislation and turns it into a federal ceiling, preventing state lawmakers from addressing emerging AI harms in an era of fast-moving technology,” Carson said. “Over the past two decades, state lawmakers have proven to be a backstop for tech accountability, fighting for families and communities even as Congress has stalled.”
Brendan Steinhauser of the Alliance for Secure AI offered a similarly mixed verdict, praising the bill’s bipartisanship and focus on catastrophic risk but concluding it “falls short of the framework necessary to keep Americans safe from the dangers of advanced AI.”
Industry Reaction: Cautiously Supportive
Tech industry groups have been more welcoming, if not without reservations. NetChoice’s Patrick Hedger expressed general support for the legislation but flagged concern about “the bill’s aggressive auditing regime and data-sharing requirements for developers that could risk entrepreneurs’ trade secrets and private records.”
This tension — between regulators wanting transparency and companies protecting proprietary systems — will likely be the central battleground as the bill moves from discussion draft to formal legislation.
Rep. Obernolte emphasized the bill’s purpose in his statement: “This discussion draft is an important step toward building a clear federal framework that promotes innovation, protects Americans from emerging risks, and ensures the United States continues to lead the world in AI. We are releasing this draft to hear from stakeholders, experts, and the public so we can strengthen the legislation before it is formally introduced.”
Workforce and Education Provisions
Beyond safety regulations, the bill includes a significant workforce component that could help soften political opposition. It would:
- Fund grants for developing AI-literacy curriculum in schools.
- Provide scholarships for students studying AI.
- Create an AI Workforce Research Hub within the Labor Department to study how AI is affecting employment.
- Require enhanced data collection on layoffs related to AI-driven workforce changes.
These provisions appear aimed at addressing concerns from labor groups and Democrats who fear AI-driven job displacement, and they give Trahan — the bill’s Democratic co-sponsor — political cover on the state preemption issue.
The Bigger Picture
The Great American AI Act arrives at an inflection point. In the past week alone, Anthropic confidentially filed an S-1 with the SEC targeting an IPO valuation near $1 trillion, ChatGPT launched its most significant memory upgrade in two years, and the White House issued new executive guidance on AI reviews. Washington is finally catching up to Silicon Valley — at least legislatively.
Whether the bill passes in its current form is uncertain. The preemption clause will face fierce opposition from state-rights advocates and AI safety groups, and the auditing requirements may draw pushback from the very companies the bill targets. As a discussion draft, the legislation is explicitly designed to invite revision.
But the message from Congress is clear: the era of entirely voluntary AI governance is ending. The question is no longer whether the federal government will regulate frontier AI — it’s how much room states will be left to act, and whether the rules Congress sets will be strong enough to matter.
*Sources: Roll Call, Americans for Responsible Innovation, Alliance for Secure AI, NetChoice, Rep. Obernolte’s Office, Rep. Trahan’s Office*
