Key Takeaways
Reps. Jay Obernolte and Lori Trahan introduced the bipartisan FRONTIER Act on July 23, creating the first federal oversight framework for the most advanced AI models.
The bill sets tiered rules by developer size, requiring model cards, risk-management frameworks, independent audits, and incident reporting from large frontier developers.
A single national standard would override state AI laws, the provision drawing the most debate as the bill enters a crowded fall calendar.
A bipartisan group of House lawmakers introduced the FRONTIER Act on July 23, a bill that would build the first federal framework for overseeing the most advanced artificial intelligence systems. Reps. Jay Obernolte, a California Republican, and Lori Trahan, a Massachusetts Democrat, led the introduction, joined by four other members from both parties.
FRONTIER stands for Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting. The bill scales its requirements to the size of the developer. It defines a frontier model by a very high compute threshold, more than 10^26 operations, and reserves the strictest tier for large developers, companies with more than $50 million in revenue and $1 billion in AI development spending, and very large developers above $5 billion in revenue and $10 billion in spending.
Under the bill, those companies would have to publish model cards, adopt risk-management frameworks, submit to independent audits, and report serious safety incidents to the government on a set timeline. The point is to make the biggest labs show their work on catastrophic risk rather than take their assurances on faith.
The most contested piece is structural. The FRONTIER Act would set one national standard and keep states from writing their own AI rules, which supporters say gives developers clear and consistent requirements and critics say strips states of the power to address AI harms their residents feel first. The fight over who writes the rules has become the fight over whether the bill passes.
Congress has moved slowly on AI while the models kept shipping, and a transparency-and-audit regime is a different approach than the enforcement fights that produced Anthropic's $1.5 billion copyright settlement or the White House accusation that Moonshot distilled a US model. Whether independent audits and public incident reports can keep pace with systems that change every few months is the question the bill leaves open. Worth watching.
People Also Ask
What is the FRONTIER Act?
The FRONTIER Act is a bipartisan House bill introduced July 23, 2026 that would create federal oversight for the most advanced AI models, requiring large developers to publish risk frameworks, submit to independent audits, and report serious safety incidents.
Who introduced the FRONTIER Act?
Reps. Jay Obernolte, a California Republican, and Lori Trahan, a Massachusetts Democrat, led the bill, joined by Reps. Scott Franklin, Scott Peters, Erin Houchin, and Suhas Subramanyam.
Which AI companies would the FRONTIER Act cover?
It targets frontier-model developers by size, with the strictest rules for large developers above $50 million in revenue and $1 billion in AI spending, and very large developers above $5 billion in revenue and $10 billion in spending.
Why is the FRONTIER Act controversial?
Its single national standard would preempt state AI laws. Supporters call that clarity for developers, and critics call it a loss of states' ability to address local AI harms.