OpenAI backs a provision of the FRONTIER Act, a bipartisan House bill, that would require top AI companies to embed outside evaluators to ensure model safety

Artificial intelligence giant OpenAI is supporting a bipartisan House proposal that would require top AI companies to embed outside evaluators to ensure their models are safe.

Politico 2026-09-15 Brendan Bordelon

Context & Ripple Effects

OpenAI had already accepted government access to major models through the US AI Safety Institute evaluation arrangement in 2024, and in May backed state and federal measures on transparency and incident reporting. Its position on the FRONTIER Act extends that safety posture from voluntary or agency-facing review toward a statutory outside-evaluator model.

The House proposal was introduced in June with risk-management requirements for leading developers and provisions overriding some state AI laws. OpenAI's endorsement gives the bipartisan House framework support from a company that would be among those subject to its requirements.

First-order effects

OpenAI aligns with a FRONTIER Act requirement that top AI companies embed independent evaluators, making external review a legislative position rather than solely an internal or government-access practice.
If enacted, covered AI developers would have to build outside evaluation into their model-safety processes, creating a formal compliance obligation around testing and risk assessment.

Second-order effects

Anthropic, Google DeepMind, and other leading developers face greater pressure to state whether they support comparable independent-review requirements, especially after OpenAI said it had worked with Anthropic and Google on safety measures.
Specialist evaluators gain a clearer route into frontier-model governance, while companies must reconcile their internal safety work with review by outside parties.

Third-order effects

The proposal points toward frontier AI oversight organized around auditable, repeatable assurance processes rather than voluntary company commitments alone.
If lawmakers make evaluator access a condition for leading-model development, safety governance may consolidate around a small set of recognized assessment practices and institutions.

The trend: Frontier AI policy is moving from voluntary safety pledges and agency access toward externally verifiable oversight embedded in law.

Discussion

@dkthomp Derek Thompson on x
I feel like I'm taking crazy pills with how many smart people I read and respect are saying stuff like this. The idea that the AI labs are suddenly, only now, just in September of 2026, advocating for AI safety, and that they just came around to this position bc of sudden financi…
@jawillick Jason Willick on x
Maybe they think “outside evaluators” can be a liability shield.
@mattplatkin Matt Platkin on x
1. Why do we need their blessing? They didn't ask for our permission to recklessly create technology that, by their own admission (!), can kill us all. 2. Can we try relying on expertise of people not seeking trillions of dollars for the thing we are trying to regulate? 3. An ear…
@thelastrefuge2 @thelastrefuge2 on x
The Dept of Federal Cyber Compliance. Gee, what could possibly go wrong?
@kevinbankston Kevin Bankston on x
Notable...Wonder what changes they are asking for ("specifically talked about how we think about IVO and made clear that we can support that," emphasis mine), wonder where Anthropic is on the current text, etc.
@emma_dumain Emma Dumain on x
Potentially a big deal for a bipartisan bill that's been slow to pick up traction, despite being the only comprehensive AI safety framework ready for committee action on the Hill
@zachary Zach Warmbrodt on x
NEW: OpenAI is backing a bipartisan House proposal that would require top AI companies to embed outside evaluators to ensure their models are safe. @brendanbordelon https://www.politico.com/...

Chronicles