Reps. Trahan and Obernolte introduce the bipartisan FRONTIER Act: licensed independent auditors, incident reporting and state preemption for frontier AI
On July 23, 2026 Reps. Lori Trahan (D-MA) and Jay Obernolte (R-CA) introduced the FRONTIER Act (H.R. 9925), a tiered federal framework for frontier AI developers: model cards, risk-management frameworks, safety-incident reporting and, for the largest developers, ongoing assessment by licensed independent verification organizations. It grew out of their June "Great American AI Act" discussion draft, which included a three-year preemption of state laws on model development. In September OpenAI's Chris Lehane backed parts of it.
Key facts
- Full name: Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting (FRONTIER) Act; H.R. 9925, 119th Congress (CBS News)
- Introduced July 23, 2026, the same day as the Lieu–Moran AI Kill Switch Act; co-sponsors Scott Peters (D-CA), Scott Franklin (R-FL), Suhas Subramanyam (D-VA), Erin Houchin (R-IN)
- Tiered by developer size: model cards, risk-management frameworks, independent audits, incident reporting, ongoing assessments; 'very large' developers retain a licensed independent verification organization
- Government licenses competing verifiers and can revoke a license when a verifier's work does not hold up
- Predecessor: the June 2026 'Great American AI Act' discussion draft (FPF analysis, June 9): $50M revenue threshold for 'frontier developers', 15-day incident reporting (24 h if imminent danger), audits every six months, whistleblower protections, three-year preemption of state laws specifically regulating model development
- Sept 16, 2026: OpenAI's chief global affairs officer Chris Lehane supported portions of the bill at a Washington roundtable, without endorsing all of it (CBS News)
What happened
The FRONTIER Act is the main bipartisan House framework for frontier AI. Instead of a government agency testing models, it relies on licensed private verifiers who check that the largest developers' safety protocols keep catastrophic risk at acceptable levels, plus public transparency reports and incident reporting. A revised version was later released with broader stakeholder support (Trahan and Obernolte press materials); the exact date of that revision was not verified.
Why it matters
It is the House counterpart to the Senate's duty-of-care talks and the industry-favoured model (third-party audits plus preemption of state laws such as California SB 53). OpenAI's partial endorsement made it the bill most labs could live with.
Changelog
- 2026-10-03: created
Related events
- Reps. Lieu and Moran introduce the bipartisan AI Kill Switch Act (H.R. 9917) after the OpenAI–Hugging Face incident ★★★
- Thune, Cruz and Klobuchar negotiate a Senate bill imposing a 'duty of care' on frontier AI developers and letting the government block unsafe model releases ★★★★
- Ted Cruz blocks the Warner–Schatz–Kim AI Risk Management and Security Act, which would give a federal AI Safety Board 45 days with frontier models before release ★★★
- California enacts SB 53, the first US frontier AI transparency law ★★★
Sources (5)
- officialRep. Trahan: Trahan, Obernolte introduce bipartisan FRONTIER Act to strengthen oversight of advanced AI
- pressCBS News: OpenAI backs measure that would require independent audits of AI models
- pressFuture of Privacy Forum: Frontier AI goes federal: how the Great American AI Act compares to state laws
- pressFathom applauds introduction of the FRONTIER Act
- discussionLessWrong: Congress moves at tech pace: the FRONTIER Act
id: 2026-07-23-frontier-act-trahan-obernolte · updated 2026-10-03 · open in the interactive timeline