Bipartisan Frontier AI Safety Bill Introduced in Congress to Restrict Catastrophic-Risk Deployments
Summary
- • Reps. Obernolte (R-CA) and Trahan (D-MA) introduced a bipartisan bill targeting the most powerful AI models.
- • The bill would restrict frontier AI deployment when models present imminent catastrophic risks.
- • Legislation is narrowed from the lawmakers' earlier comprehensive AI regulatory draft.
- • Focus is specifically on extreme-risk scenarios rather than broad AI governance.
Details
Frontier AI deployment restrictions
Bill would restrict deployment of the most powerful AI models when they present imminent catastrophic risks to the public.
Bipartisan sponsorship
Introduced by Rep. Jay Obernolte (R-CA) and Rep. Lori Trahan (D-MA) — one Republican, one Democrat — signaling cross-party support for targeted AI safety rules.
Derived from comprehensive AI draft
The bill draws from portions of Obernolte and Trahan's earlier draft comprehensive AI regulatory proposal, with changes specifically addressing the most extreme risk scenarios.
Narrow scope by design
Rather than broad AI regulation, the bill focuses on catastrophic-risk scenarios — a deliberate strategy to increase legislative viability while still addressing the highest-stakes AI safety concerns.
Policy details distilled from Punchbowl News reporting on the bill's introduction.
What This Means
The Obernolte-Trahan frontier AI bill marks a notable bipartisan step toward U.S. federal AI safety legislation, targeting the narrow but high-stakes scenario of frontier models posing imminent catastrophic risks. By deliberately narrowing scope from their earlier comprehensive regulatory draft, the lawmakers appear to be pursuing a more achievable legislative path in a divided Congress. If passed, this would establish the first federal legal constraints on deploying the most powerful AI systems — with direct implications for leading AI labs developing frontier models. The bill's high "imminent catastrophic" threshold suggests it is designed to avoid stifling mainstream AI development while providing a legal backstop for worst-case scenarios.