A 269-page draft bill, titled the Great American AI Act, released in early June by Rep. Jay Obernolte and Rep. Lori Trahan, would govern federal AI research funding, cybersecurity, classroom curricula, teacher training, workforce programs, and safety rules for large models, and it is being structured as separate bills for committee review.

There is mounting political pressure to advance a floor vote and safety measures, with bipartisan talks and calls for international collaboration on safety metrics and testing protocols, including engagement with China.

Industry voices acknowledge widespread AI adoption, noting OpenAI’s user growth and the emphasis on workforce training, productivity measurement, and ensuring broad economic benefits while addressing regional disparities.

Advocates push for transparency, external auditing, and safety, with lawmakers urging disclosure, third-party audits, and actions to mitigate catastrophic risk from cyber threats and misaligned AI objectives.

A core dispute centers on a three-year preemption of state AI laws under Title I, likely overriding state rules on model development, transparency of training data, privacy, and civil rights, while preserving state use regulations.

The FRONTIER Act would impose obligations on frontier models, including publishing model information, risk management plans, safety incident reporting, and independent verification, while lighter requirements would apply to users of AI tools built by others.

The FRONTIER Act, aimed at the most powerful AI systems, sits with the Energy and Commerce Committee and could hear testimony in November, with a possible floor vote in December if paired with other legislation.

Eleven bipartisan AI-related bills cleared the House Science Committee in July, but none have reached the House floor yet, as the package derived from the Great American AI Act was divided into 11 separate bills across eight committees.