AI regulation in the United States has entered a defining new chapter. On June 4, 2026, Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a 269-page bipartisan discussion draft titled the “Great American Artificial Intelligence Act” — the most comprehensive federal AI regulation proposal the US Congress has ever produced. The bill would create binding obligations for large AI developers, preempt a wave of conflicting state laws for three years, and establish a national governance framework that could shape the global AI industry for decades.
Key Takeaways
- The Great American AI Act is a 269-page bipartisan bill introduced June 4, 2026 by Reps. Obernolte and Trahan.
- The bill would preempt state AI regulation for three years to create a single federal AI regulation framework.
- It would require frontier AI developers — companies with $500M+ in annual revenue that trained a frontier model — to disclose model details and undergo third-party audits.
- The bill covers four areas: Frontier AI Governance, Workforce, Cybersecurity, and Research & International Cooperation.
- The EU AI Act begins applying on August 2, 2026 — adding urgency to the US push for its own federal AI regulation standard.
What Happened?
AI regulation in the US has until now consisted of a patchwork of executive orders, agency guidance, and state-level laws — none of which add up to a coherent national framework. The Great American AI Act aims to change that. Released as a discussion draft on June 4, 2026, the bill is bipartisan in sponsorship and sweeping in scope. It is supported by four additional members of Congress across both parties: Scott Franklin (R-FL), Suhas Subramanyam (D-VA), Erin Houchin (R-IN), and Scott Peters (D-CA).
The bill is structured around four titles. The first — Frontier AI Governance — is the most consequential. It would apply to “large frontier developers,” defined as companies with $500 million or more in annual revenue that have trained a frontier AI model. These companies would face three core obligations: disclosing detailed technical information about their models, submitting to audits conducted by designated Independent Verification Organizations (IVOs), and refraining from retaliating against employees who raise safety concerns internally or externally.
The second title addresses the workforce implications of AI deployment — a politically sensitive area that has been central to union negotiations across several major industries. The third title covers cybersecurity requirements for AI systems used in critical infrastructure. The fourth addresses research and development priorities and international cooperation, including provisions aimed at keeping the US competitive relative to China’s AI programme.
The state preemption provision has drawn the most immediate controversy. The bill would pause all state-level AI regulation specifically targeting model development for three years, with a sunset clause. Colorado, which passed a comprehensive AI consumer protection law in 2025, would see its rules preempted under the proposal. Consumer advocacy groups including Public Citizen have criticised the provision as stripping state-level protections from workers, consumers, and children before a federal standard is fully in place.
Why It Matters
AI regulation has moved from a peripheral policy debate to a central legislative priority in Washington. The Great American AI Act matters for several interconnected reasons. First, it would be the first federal law in US history that directly governs the development of AI models — not just their deployment in specific sectors, but the models themselves. That is a significant jurisdictional step that would permanently reshape the relationship between government and the AI industry.
Second, the timing is urgent. The EU AI Act begins full application on August 2, 2026. European companies operating under that framework will face requirements around transparency, audit, and risk classification that their US competitors currently do not. A federal US AI regulation standard would level the playing field for American companies doing business internationally and reduce the risk of the US defaulting to European rules by default — a dynamic that trade analysts have called regulatory arbitrage.
Third, the discussion draft format is intentional. Rather than introducing a polished bill for an immediate vote, Obernolte and Trahan have released the draft for public comment and stakeholder input. This approach is designed to build coalition support from the AI industry, civil society, and state governments before formal introduction. Legal analysts at DLA Piper have noted that the draft’s structure reflects careful drafting to avoid the constitutional challenges that have plagued other federal preemption efforts in the technology space.
The rapid acceleration of AI adoption globally makes the regulatory gap increasingly risky. Without a federal framework, enforcement of AI rules in the US would continue to fall to the FTC, sector regulators, and state attorneys general — a fragmented approach that creates inconsistent outcomes and higher compliance costs for companies operating nationally.
Expert Analysis
Legal and policy experts have given the Great American AI Act a cautiously positive reception, while flagging several provisions as likely to require revision before passage. The Independent Verification Organization model — where third-party auditors assess frontier models for safety and compliance — is seen as a workable analogue to financial auditing frameworks, but experts note that the technical capacity to audit large AI models at scale does not yet exist in the private sector. Building out that infrastructure would take time and investment that the bill’s timeline may not fully account for.
The whistleblower protection provisions have drawn more unambiguous praise. Protections for employees who raise safety concerns internally before going to regulators are seen as essential to the effectiveness of any AI oversight regime. Without them, the information asymmetry between large AI developers and regulators would remain nearly impossible to bridge — regulators simply cannot independently assess the internal state of a frontier model without cooperation from people inside the developing organisation.
The $500 million revenue threshold for “large frontier developers” has been questioned as potentially too high for the current market. Several companies that have trained commercially deployed frontier models — including some that have been acquired by larger firms — would not meet this threshold, creating regulatory gaps. Conversely, critics from the industry argue that the threshold is arbitrary and that a capability-based definition would be more appropriate, focusing on the properties of the model rather than the financial size of the company that built it.
The Cato Institute’s analysis of the draft points out that the bill’s preemption mechanism, while designed to reduce regulatory fragmentation, could create a vacuum during the three-year moratorium if Congress does not pass a final version in that window. The draft itself does not specify what happens when the preemption period ends without final legislation in place — a gap that state attorneys general have already flagged in public comments.
Industry Impact
The AI industry’s reaction to the Great American AI Act has been divided along predictable lines. Large AI developers — including OpenAI, Google DeepMind, and Anthropic — have broadly welcomed the federal AI regulation framework as preferable to a patchwork of state laws that would create compliance complexity without commensurate safety benefits. Industry associations have called the bill a “serious starting point” and indicated support for the audit and disclosure architecture, while pushing back on specific provisions around preemption timelines and revenue thresholds.
Smaller AI companies and open-source developers have expressed concern that the compliance requirements would create barriers to entry that benefit incumbents. The cost of a third-party IVO audit, conducted according to standards that do not yet exist, could be significant — particularly for startups that compete with the large frontier developers the bill targets. Several open-source advocates have argued that the bill’s definition of “frontier model” needs to be clarified to explicitly exclude open-source models released without commercial intent.
The workforce title has drawn the most attention from organised labour. Unions representing workers in industries with significant AI adoption — including logistics, healthcare administration, and financial services — have called for stronger provisions around worker notification and transition assistance. The bill’s current language on workforce implications is seen as a framework for further negotiation rather than a final settlement.
On the international dimension, the bill’s provisions on cooperation align with ongoing discussions at the G7 and the OECD on AI governance standards. The intensifying global AI race between the US and China has given the international cooperation title particular urgency — both as a mechanism for sharing safety research with allies and as a tool for coordinating export controls on the most capable AI systems.
What’s Next for AI Regulation
The discussion draft is open for public comment through the summer of 2026. Obernolte and Trahan have indicated they will hold roundtable sessions with AI developers, civil society groups, state regulators, and international partners before formally introducing the bill for a committee vote. The timeline for a final legislative vote is not yet set, but the sponsors have expressed hope for passage before the end of the current Congressional session.
Several parallel tracks are running simultaneously. The White House AI Innovation and Security order issued in June 2026 directs federal agencies to accelerate AI adoption within a framework of defined safety standards — a parallel process that will interact with the congressional effort in ways that are still being worked out. The National Institute of Standards and Technology is developing updated AI risk management frameworks that could eventually become the technical backbone for the IVO audit standards the bill envisions.
State-level responses are also forming. Colorado, whose 2025 AI consumer protection law would be preempted under the bill, has signalled that it will seek to shape the federal standard rather than simply accept preemption. Several other states with pending AI legislation have paused their processes to see how the federal bill develops — a practical acknowledgement that a single national framework, if it materialises, would be preferable to a 50-state compliance landscape for companies and consumers alike.
The Great American AI Act is not guaranteed to pass in its current form. But it represents the clearest signal yet that the US government is moving toward a definitive federal AI regulation framework — one that could arrive just in time to prevent the US from falling permanently behind the regulatory standards being set by Europe.
Frequently Asked Questions
What is the Great American AI Act?
The Great American AI Act is a 269-page bipartisan federal AI regulation bill introduced as a discussion draft on June 4, 2026 by Representatives Obernolte and Trahan. It would create binding obligations for large frontier AI developers, preempt state AI laws for three years, and establish an audit and disclosure framework overseen by Independent Verification Organizations.
Who does the AI regulation bill apply to?
The bill targets “large frontier developers” — companies with $500 million or more in annual revenue that have trained a frontier AI model. Companies below this threshold are not covered by the bill’s core obligations.
What would the Great American AI Act require companies to do?
Covered companies would need to disclose technical information about their frontier models, submit to third-party safety audits conducted by designated Independent Verification Organizations, and protect employees who raise safety concerns from retaliation.
Why does the bill preempt state AI laws?
The bill’s sponsors argue that a patchwork of conflicting state AI regulation creates compliance complexity without improving safety outcomes. The three-year preemption is designed to give Congress time to establish a unified federal framework. Critics argue it removes existing consumer protections before replacements are in place.
When could the Great American AI Act become law?
The discussion draft is open for public comment through summer 2026. A formal committee vote and final legislative timeline have not been set. The sponsors aim for passage before the end of the current Congressional session.
Conclusion
AI regulation in the United States is at an inflection point. The Great American AI Act is the most ambitious federal attempt yet to establish governance rules for the development of large AI models. Its passage would fundamentally alter the relationship between government and the AI industry — creating oversight mechanisms that proponents argue are essential for safety and that critics warn could slow innovation and entrench incumbents.
What is not in doubt is that the question of AI regulation is no longer theoretical. With the EU AI Act taking effect in August 2026 and the global AI race intensifying, the US faces a choice between shaping global AI governance standards or accepting rules set elsewhere. The Great American AI Act is Congress’s answer to that choice — imperfect, contested, but unmistakably serious.
Sources
– Rep. Obernolte Office: Discussion Draft of the Great American AI Act
– Roll Call: Bipartisan AI draft proposes three-year preemption of state laws
– DLA Piper: Unpacking the Great American AI Act
– Public Citizen: Obernolte-Trahan Bill Strips States Authority to Protect Consumers
This article is for informational purposes only. AI developments mentioned are based on publicly available sources and do not constitute endorsement of any product or technology.
