US House Proposes 'Great American AI Act' to Create Federal Framework, Preempt State Laws for Three Years

A bipartisan coalition in the U.S. House of Representatives has officially unveiled the 'Great American Artificial Intelligence Act' (GAAIA), a sweeping 269-page discussion draft that represents Congress's most ambitious attempt yet to establish a unified federal framework for AI governance. Released jointly by Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA), the comprehensive proposal seeks to balance the rapid pace of technological innovation with necessary national security and consumer safeguards. By releasing the text as a discussion draft, lawmakers are actively soliciting feedback from industry experts, civil rights advocates, and the public before formally introducing the legislation to the House floor.[3]

The legislation arrives at a critical juncture for the American technology sector, which is currently navigating a rapidly expanding and increasingly complex patchwork of state-level AI regulations. By proposing a centralized federal standard, the bill aims to solidify United States leadership in artificial intelligence while providing developers with a predictable, unified regulatory environment. However, its specific approach to achieving that national uniformity has already ignited a fierce, high-stakes debate over the balance of power between state legislatures and federal authorities, drawing battle lines that cross traditional partisan divides.[1][2]

The most consequential and politically contentious provision of the GAAIA is Section 3, which mandates a strict three-year federal preemption of state and local laws that specifically regulate the development of artificial intelligence models. If enacted into law, this clause would effectively freeze state-level efforts to impose mandatory safety testing, algorithmic auditing, or transparency requirements on the training and fine-tuning of frontier AI systems. This temporary moratorium is designed to give the federal government exclusive jurisdiction to establish baseline testing and oversight frameworks without developers having to navigate conflicting local mandates.[3]

Proponents of the preemption clause argue that a fragmented regulatory landscape is fundamentally untenable for a digital technology that inherently crosses state and national lines. Representative Obernolte has publicly emphasized that the most advanced AI systems are built in a single state but deployed nationwide, shaping jobs, consumers, and public safety everywhere, making zip-code-dependent protections inherently insufficient. Industry advocates and tech coalitions echo this sentiment strongly, warning that conflicting state mandates could impose crippling compliance costs, stifle domestic innovation, and ultimately cede critical geopolitical ground to international competitors.[1][3]

Conversely, the preemption clause has drawn sharp, immediate opposition from state officials and civil society organizations who view it as an overreach. A bipartisan coalition of more than 200 state lawmakers representing 42 states recently issued a joint letter urging Congress to reject the provision, arguing that state legislatures have historically served as the primary and most agile backstop for tech accountability. Organizations like Americans for Responsible Innovation warn that the bill creates an artificial 'federal ceiling' that would prevent local governments from rapidly addressing emerging algorithmic harms to children, workers, and marginalized communities while Congress moves at a slower pace.[1]

Despite the sweeping nature of the preemption debate, legal analysts point out that the bill's jurisdictional scope is carefully circumscribed to avoid total federal control. The legislation specifically targets the underlying development phase of AI models—such as determining training objectives and modifying neural weights—leaving state laws that govern the downstream deployment and commercial use of AI entirely intact. This critical distinction means that state-level regulations concerning employment discrimination, consumer privacy, healthcare liability, and common-law fraud would remain fully enforceable against companies that use AI in their daily operations.[3]

Beyond the jurisdictional tug-of-war, the GAAIA introduces robust, mandatory oversight mechanisms specifically tailored for the industry's largest and most capable players. The legislation imposes new transparency and auditing requirements on 'frontier' AI developers, which the bill defines as companies generating over $500 million in annual gross revenue. These well-resourced corporations would be legally mandated to publish comprehensive safety frameworks, report critical security incidents to the government, and submit their most advanced models to independent third-party evaluations before public release.[3]

To manage and enforce this new regulatory apparatus, the bill formally codifies the Center for AI Standards and Innovation (CAISI) as a permanent entity within the Department of Commerce. Originally established under the Biden administration as the AI Safety Institute and recently rebranded, CAISI would be granted statutory authority and a dedicated director. The center would be tasked with developing voluntary industry guidelines, creating standardized evaluation tools, and continuously monitoring the technology's progress against established national security and public safety benchmarks.[1][3]

The legislation also directly addresses the profound economic and societal shifts anticipated from the widespread corporate adoption of artificial intelligence. It directs the Bureau of Labor Statistics and the Census Bureau to fundamentally revise federal surveys to accurately track AI integration and worker displacement across the national workforce. Furthermore, the bill authorizes the National Science Foundation to establish targeted grants focused on AI education, workforce development, and reskilling, signaling a significant federal commitment to mitigating the inevitable labor market disruptions.

In a concerted bid to democratize access to the massive computational power required for modern AI research, the GAAIA would permanently establish the National Artificial Intelligence Research Resource (NAIRR). By transitioning NAIRR from a temporary pilot program to a fully funded statutory resource, the bill aims to provide universities, non-profit researchers, and small startups with the hyperscale computing infrastructure and vast datasets that are currently dominated by a handful of trillion-dollar tech giants.

Cybersecurity forms another core, non-negotiable pillar of the proposed federal framework. The draft legislation explicitly reauthorizes the Cybersecurity Information Sharing Act of 2015, extending crucial liability protections to incentivize private sector tech companies to share urgent cyber threat data with the federal government. It also creates newly funded grant programs administered through the Cybersecurity and Infrastructure Security Agency (CISA), which are specifically designed to harden the defenses of open-source AI software against malicious exploitation.[1][3]

Recognizing the opaque internal dynamics of frontier AI laboratories, the bill introduces explicit, legally binding whistleblower protections for tech workers. Section 113 of the draft shields employees and independent contractors from corporate retaliation if they report violations of federal law related to the development, deployment, or operation of AI systems. This provision directly addresses growing concerns within the AI safety community about the lack of transparency and the intense pressure applied to researchers at leading commercial labs.[3]

While the GAAIA is currently circulating only as a discussion draft, its individual components are already generating significant legislative momentum on Capitol Hill. The House Science, Space, and Technology Committee recently advanced 10 bipartisan AI bills in a single markup session that mirror the non-regulatory research and security provisions of the GAAIA. This parallel movement indicates that Congress is actively laying the technical and institutional groundwork for more substantive AI oversight, even as the broader regulatory debate continues.

The path forward for the comprehensive package, however, remains highly complex and politically fraught. Passing a unified bill that includes the controversial state preemption clause will require delicate, high-stakes negotiations between federal standardization advocates and staunch state regulatory defenders. As Congress continues to solicit feedback from a wide array of stakeholders, the Great American AI Act stands as the most detailed and ambitious blueprint yet for how the United States intends to govern the defining technology of the twenty-first century.[2]