House Draft Targets Frontier Labs with Audits While Blocking State AI Laws
On June 5, 2026, a newly proposed bipartisan bill aimed at creating a comprehensive federal framework for regulating frontier artificial intelligence has sparked immediate backlash following its release. Critics have expressed grave concerns about a specific provision designed to prevent any state-level regulations on AI model development for a period of three years. The bill, titled the Great American Artificial Intelligence Act of 2026, was introduced by Representatives Jay Obernolte, a Republican from California, and Lori Trahan, a Democrat from Massachusetts. It garnered support from four co-sponsors, including Representatives Suhas Subramanyam, Scott Franklin, Scott Peters, and Erin Houchin, showcasing a rare moment of bipartisan collaboration over a contentious issue.
The unveiling of the draft legislation came just two days subsequent to a significant executive order signed by President Trump, which directed federal agencies to conduct voluntary reviews of frontier AI models. This executive action underscores the administration’s preference for minimal regulation, indicating a determination to establish a federal framework swiftly to prevent states from independently advancing their own regulations on AI. As Congress prepares to enter the final months of its session before the anticipated August recess, lawmakers are increasingly aware of the scrutiny they will face as the November midterm elections loom closer.
The 269-page bill outlines several key requirements for powerful AI companies that are labeled as "large frontier developers." These firms, including industry giants like OpenAI, Anthropic, and Google, will be obliged to publish safety frameworks, submit semi-annual audits conducted by independent entities, and report any critical incidents to federal regulators. Non-compliance can lead to severe financial repercussions, with penalties accumulating at a rate of up to $1 million per day.
One of the bill’s most controversial aspects is its specific prohibition against states establishing or enforcing regulations that focus on AI model development for three years. While the legislation does carve out exceptions for laws related to civil rights, labor protections, consumer privacy, and the prevention of child sexual abuse material, it nonetheless centralizes federal authority over a rapidly evolving domain while limiting state-level action.
Labor unions, civil rights organizations, and advocates for AI safety have promptly condemned this approach. Brad Carson, president of Americans for Responsible Innovation, criticized the bill’s provision, predicting it would hinder state lawmakers from tackling emerging AI-related issues in a technology landscape that is evolving at breakneck speed. Carson’s organization has launched advertising campaigns in Massachusetts aimed at persuading Representative Trahan to denounce the preemption provisions, labeling the approach as a "generational mistake."
Brendan Steinhauser, CEO of The Alliance for Secure AI, further articulated concerns, advocating for a minimum national AI standard that would match or exceed existing state regulations designed to protect children and consumer rights. Union leaders voiced strong disapproval as well, suggesting that the bill effectively serves the interests of the AI industry at the potential expense of public safety. Randi Weingarten, president of the American Federation of Teachers, characterized the bill as a handout to the AI sector, while the AFL-CIO rigorously condemned any attempts to restrict states from safeguarding the welfare of workers.
Critics also pointed out discrepancies within the bill. An official FAQ document from Representative Trahan’s office attempted to clarify legal thresholds between "preempted" and "federalized" laws; however, the bill’s text does not make any such distinctions. Nicholas Garcia, senior policy counsel at Public Knowledge, raised concerns that the broad language of the bill might encompass far-reaching implications, potentially capturing legislation unrelated to core AI frameworks under its scope.
As the discourse continues, Garcia suggested that a more thoughtful approach to preemption could enhance the bill’s viability, arguing for a "one-to-one preemption" model that would allow states to maintain autonomy over matters not explicitly addressed by federal regulations. Cobun Zweifel-Keegan, managing director at the International Association of Privacy Professionals, agreed, affirming that the preemption language appears overly broad, which could restrict state-level oversight even when the issues at hand differ fundamentally from the bill’s central focus.
Co-sponsor Erin Houchin defended the motivations behind the bill, asserting the necessity of establishing a national standard in order for America to maintain its leadership in the AI sector. She cautioned against the pitfalls of creating a patchwork of varying state laws that could leave the nation at a competitive disadvantage, particularly when juxtaposing the United States with countries like China.
Despite concerns about the volume of state legislation, Garcia argued that only a handful of states have instituted significant AI regulations, many of which coexist harmoniously. The current legislative draft primarily applies to large AI developers generating over $500 million in annual revenue while defining specific criteria for frontier models that necessitate more thorough scrutiny.
The proposed framework’s structure involves three interconnected components mandating large developers to submit a frontier AI framework that provides technical protocols for addressing "catastrophic risk" scenarios. These frameworks will also require transparency reports prior to deploying any new or modified AI models, thus encouraging a greater degree of accountability within the industry.
The draft legislation aims to establish the Center for AI Standards and Innovation, allocating $100 million annually for fiscal years 2027 to 2029. This center would play a critical role in developing guidelines and security standards for AI technologies while overseeing the licensing of independent verification organizations tasked with auditing AI operations.
As discussions about this legislation unfold, subtle shifts in its language and structure could address some of the prevailing critiques without sacrificing core objectives. The complexity of the issues at hand demands careful consideration, emphasizing the need for collaborative approaches that acknowledge both the urgency of regulation and the importance of state-level engagement in managing AI technologies.