Bipartisan FRONTIER Act Targets AI Catastrophic Risks With Federal Oversight
Reps. Obernolte and Trahan introduce legislation requiring safety audits and emergency shutdown authority before 2026 midterms.
Two House lawmakers introduced the FRONTIER Act on July 23, 2026, establishing the first comprehensive federal framework for regulating frontier AI models. The bipartisan bill, sponsored by Rep. Jay Obernolte (R-CA) and Rep. Lori Trahan (D-MA) with four cosponsors, mandates safety reporting, independent audits, and grants the Secretary of Commerce emergency authority to halt dangerous AI systems.
Emergency Shutdown Authority
The legislation empowers the Commerce Secretary to suspend or restrict development, deployment, or internal use of frontier models upon a written finding of "imminent catastrophic risk." The bill defines this threshold precisely: 50 or more deaths or serious injuries, or over $1 billion in property damage resulting from specific model behaviors. This shifts enforcement from voluntary corporate self-regulation to federal oversight.
Mandatory Safety Reporting
Large AI developers must publish safety reports, report critical safety incidents, and undergo independent third-party audits. Federal preemption is narrowly tailored to three areas: catastrophic-risk transparency, third-party auditing and independent verification, and incident reporting. States retain authority over laws protecting minors, generally applicable laws, deployment and use-based regulation, and state procurement decisions.
Political Context and Timing
The July 2026 autonomous AI breach involving OpenAI and Hugging Face catalyzed the legislation's urgency. "There is real urgency. There's earth-shattering news practically every week, and that's only going to get more frequent and more serious as these models get more powerful," Trahan said.
The 119th Congress has struggled to pass comprehensive AI legislation amid ideological divides over government intervention in private sector innovation. To date, the only significant AI-related bill enacted is the TAKE IT DOWN Act, which criminalizes AI-generated deepfakes and non-consensual intimate imagery.
Obernolte framed the bill as a measured approach. "I don't think anyone's saying that we shouldn't have any regulation of AI," he said. Sponsors are pushing for committee action before the November 2026 midterms, though Energy and Commerce Chair Brett Guthrie remains noncommittal on the timeline.
What's at Stake
If passed, the FRONTIER Act would establish legal precedent for government-mandated safety audits and shutdown authority for AI systems. The legislation balances existential risk management with innovation preservation, creating a potential blueprint for how the United States regulates catastrophic risks from frontier AI while maintaining state-level protections.