OpenAI proposes a federal framework built on state frontier-safety laws

What the blueprint actually asks for

OpenAI has published a blueprint arguing that the U.S. should build a durable federal framework for governing increasingly capable AI systems. It is a policy proposal, not a product or a model release, and it is aimed squarely at the federal government.

The document lays out three moves. First, build a national framework that draws on the consensus already visible in state frontier-safety laws. Second, strengthen CAISI as the federal government's primary institution for frontier AI safety. Third, mobilize a broader resilience plan across government to handle the national security and public safety challenges posed by frontier AI.

Each of those parts describes a different layer: legislation, an institution to run it, and a whole-of-government response beyond any single agency. The blueprint treats all three as necessary together rather than as alternatives.

Building on statutes that already exist

The most concrete choice in the blueprint is what it points to as precedent. Rather than sketch an ideal regime, OpenAI names specific state laws already on the books: California's SB 53, New York's RAISE Act, and Illinois's SB 315.

States have started developing harmonized approaches to frontier AI governance, including California's SB 53, New York's RAISE Act, and Illinois's SB 315.Montana Labs

The company also credits the White House executive order on Promoting Advanced Artificial Intelligence Innovation and Security as a step forward. The framing is that a foundation exists and the federal government should build on it, not start over. That is a notable position for a frontier lab to take publicly, given that state-by-state rules are often described by industry as a fragmentation problem to be preempted.

Here the argument is the reverse: the state laws are treated as an emerging consensus worth codifying nationally. The blueprint wants harmonization upward rather than a clean federal slate.

CAISI as the load-bearing institution

The proposal rests on a single named body. OpenAI calls for strengthening CAISI as the federal government's primary institution for frontier AI safety, which makes the durability of the whole framework depend on the durability of that institution.

The blueprint explicitly wants a federal framework capable of evolving alongside the technology itself. That is a demanding requirement to place on any agency, and the source text does not detail what authorities, funding, or staffing CAISI would need to meet it. The proposal identifies the institution but leaves its enforcement teeth undescribed in this announcement.

The implication: a lab lobbying to lock in its own oversight

The specific implication of this blueprint is that OpenAI is actively shaping the rules it expects to be governed by, and doing so by endorsing existing state statutes rather than resisting them. A company at the frontier is asking for a federal regime and naming the institution that should run it.

For anyone building on frontier models, the signal is that the regulatory surface is being defined now, at the state level first and federally next, with a vendor helping draw the map. The blueprint's timing, released as an executive order and three state laws are already in force, suggests the window OpenAI sees is short. What the source does not answer is who audits the auditor once CAISI becomes the framework's single point of authority.