Reps. Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.) introduced legislation on Thursday to establish a federal regulatory framework for artificial intelligence. The bill is a revised version of a discussion draft released last month and aims to standardize governance for high-level AI models.
The updated text modifies previous language regarding the preemption of state laws, a point of contention for some advocates and lawmakers. Representative Obernolte stated that the new version narrows the scope of federal preemption but removes a previous three-year sunset clause. According to a bill summary, the legislation would preempt state regulations concerning catastrophic-risk transparency and third-party auditing while allowing states to maintain authority over AI deployment and child safety.
The proposed law requires developers of large-scale AI models to publish safety reports, adopt risk mitigation frameworks, and undergo independent audits. It grants the Secretary of Commerce the authority to suspend or restrict the use of models identified by auditors as posing an imminent catastrophic risk. Despite the bipartisan introduction, the bill faces a limited legislative window for floor consideration before the upcoming midterm elections.