OpenAI's latest policy move is not just another plea for federal uniformity. A sensor answer on OpenAI's own Sept. 9 post found that the company explicitly called for mandatory, capability-based national AI-safety regulation and said it would keep supporting state legislation until Congress acts [E1]. That puts the jurisdiction question in a narrower frame: whether Congress writes a national floor that still leaves state rules alive, or a preemption rule that blocks them.
Reuters gave the policy turn its immediate legislative setting. It reported that OpenAI now backs mandatory national AI-safety requirements while continuing to support state measures until Congress acts, and that it endorsed four California bills covering independent safety assessments, AI-auditor standards, youth protections and bio-threat safeguards [E2]. The operative change is the pairing of a mandatory federal standard with support for named state bills, not a general preference for one national rule.
The California status matters because the state package no longer sits in a single legislative posture. A sensor-supplied review of California Legislative Information found SB 813 and AB 1405 approved and chaptered on Sept. 9 as Chapters 179 and 178, with the bills marked inactive after chaptering [E3]. The same review found SB 1119 and AB 1864 still active and enrolled, with the governor listed as their house location after presentations on Sept. 9 and Aug. 27 [E4].
That split makes OpenAI's position more concrete than a generic national-standard slogan. Two of the endorsed measures are now state law on the evidence supplied, while two still depend on gubernatorial action, so the live question is not whether California has moved at all. It is how far a later federal statute would have to go to displace state duties that have already begun to enter the code.
The narrower supported inference is that OpenAI is trying to define a mandatory federal baseline without conceding that states must stand down before Congress acts. The company's own post and Reuters' account both preserve state action in the interim, and California's record now shows why that interim is not theoretical [E1][E2][E3][E4]. The other reading is that a call for national requirements remains a route to eventual preemption, because a federal bill could still sweep state audit, youth-safety or biosecurity rules aside.
One fact would settle that ambiguity better than any slogan: statutory language from Congress saying whether state AI-safety duties are preserved, displaced or preempted only when they conflict. Until that text exists, the enforceable clocks are state clocks, not the company's preference for a national regime. For companies asking who has authority today, the answer is narrower than the federal-versus-state fight: California has already made part of the package law, and the rest remains one signature decision away [E3][E4].