On June 5, bipartisan members of the U.S. House of Representatives unveiled a 269-page draft AI legislation, the Great American AI Act, targeting leading AI companies such as OpenAI, Google DeepMind, and Anthropic with annual revenues exceeding $500 million. The draft requires these firms to disclose their safety frameworks and undergo independent audits every six months, with violators facing fines of up to $1 million per day. The most controversial provision suspends state-level AI regulatory authority for three years. The draft also establishes whistleblower protections, increases penalties for AI-related fraud, promotes AI literacy education in K-12 schools, and extends the Cybersecurity Information Sharing Act through 2035. The draft is currently in the feedback solicitation phase, and its political prospects remain uncertain.
Author and source: AIBase
U.S. bipartisan lawmakers release 269-page draft legislation on AI.
On June 5, bipartisan members of the U.S. House of Representatives unveiled a 269-page draft legislation on artificial intelligence, aiming to comprehensively incorporate federal frameworks for safety audits of advanced AI models, whistleblower protections for employees, and research on AI’s impact on employment. The draft, titled the Great American Artificial Intelligence Act of 2026, is led by Representative Jay Obernolte, a Republican from California, and Representative Lori Trahan, a Democrat from Massachusetts, and is currently in the public feedback stage, having not yet been formally submitted for a vote.
Major frontier developers face strict constraints.
The most prominent focus of this draft is strict regulations on "large frontier developers." The bill targets companies that generated over $500 million in total revenue in the previous year and trained large frontier models—including leading labs such as OpenAI, Google DeepMind, and Anthropic. These companies must publicly disclose a "frontier AI framework" detailing how they identify and mitigate "catastrophic risks," such as assisting in the development of weapons of mass destruction, launching cyberattacks, or causing more than 50 fatalities or over $1 billion in property damage without human oversight. To ensure compliance, companies must engage independent verification agencies to audit their safety plans every six months, with violators facing civil penalties of up to $1 million per day.
State preemption clause sparks controversy
The most contentious provision in the draft is "state preemption." The bill stipulates that for three years after its enactment, state and local governments may not enact or enforce laws specifically regulating the development of AI models, though they may still regulate the use and deployment of AI systems. Supporters argue that the U.S. should not be slowed down by 50 different state rules governing development; however, opponents fear this provision would forcibly lower the strict consumer protection and child safety standards established by the states to the federal minimum.
The draft touches on the daily lives of ordinary people.
In addition to focusing on tech giants, this draft also extends its reach into the daily lives of ordinary citizens. The bill establishes whistleblower protections for employees and contractors at AI companies, safeguarding them from retaliation for reporting violations. In terms of consumer protection, it increases sentencing guidelines for financial fraud committed using AI and designates impersonating a federal official via AI as a specific offense. For students and workers, the bill requires studies on AI’s impact on employment and promotes AI literacy education in K-12 schools and university scholarship programs. Additionally, the draft plans to extend the Cybersecurity Information Sharing Act through fiscal year 2035, continuing to permit businesses to share information about cyber threats.
The political outlook for the draft remains uncertain.
Although the draft has garnered some attention within the industry, its political prospects remain uncertain. The industry group NetChoice generally supports its direction but has expressed concerns that mandatory audits and data sharing could expose trade secrets. Previously, Congress attempted to include a ten-year moratorium on state-level AI legislation in the budget bill, but it was ultimately removed by an overwhelming vote of 99 to 1 in the Senate. This draft shortens the timeframe to three years and pairs it with transparency and audit requirements; whether it can secure sufficient support through amendments and negotiations before Congress adjourns in August remains to be seen.