Two documents. Forty days apart. They point in opposite directions, and businesses operating in California have to navigate both at the same time.
On March 20, 2026, the White House released its National Policy Framework for Artificial Intelligence, a set of legislative recommendations directed at Congress. The Framework’s most consequential line calls on Congress to “preempt state AI laws that impose undue burdens to ensure a minimally burdensome national standard consistent with these recommendations.” That language is not law. It carries no binding force today. But it signals where the administration wants federal AI governance to land: a single national standard that limits states’ ability to impose their own AI requirements on businesses.
Ten days later, California moved in a different direction entirely. On March 30, 2026, Governor Gavin Newsom signed Executive Order N-5-26, directing state agencies to develop new AI vendor certification standards and reshape California’s procurement process for AI technologies. The EO applies to vendors seeking to do business with California state agencies, regardless of where those vendors are headquartered.
That last point matters. California is not only regulating its own agencies. It’s setting standards that any company wanting access to the California state market must meet. That’s procurement power used as de facto regulation, and it reaches vendors nationwide.
What’s actually binding right now
The White House Framework does not create binding legal obligations, as legal analysts at Ballard Spahr have noted. It’s a legislative roadmap. For preemption to take effect, Congress would need to pass legislation explicitly superseding state AI laws. That process is uncertain and could take years.
California’s EO N-5-26 is in effect now. California’s state agencies are reportedly directed to submit implementation recommendations within 120 days of the EO’s signing, placing an approximate deadline in late July 2026.
Why this conflict is consequential
Federal preemption of state AI laws would resolve the compliance fragmentation problem that legal teams at multi-state operators have been flagging. One federal standard beats fifty different state regimes. But it would also override state-level protections, including California’s procurement-based approach, that some view as filling gaps in federal AI governance.
The Framework is reported to recommend against creating a new dedicated federal AI regulatory body. If accurate, federal AI policy would flow through existing agencies rather than a centralized AI authority, a choice that shapes who enforces whatever federal standard eventually emerges.
The child protection provisions in the Framework, age-verification requirements for AI services likely accessed by minors, and tools for parents, represent one area where federal and California approaches may find common ground rather than conflict.
What compliance teams should watch
The immediate action item is the California EO. If your company sells technology services to California state agencies, certification standards are coming. The roughly 120-day window for agency recommendations means the shape of those standards should be clearer by late July 2026. Build that into your procurement compliance calendar.
For the federal preemption question: the Framework gives Congress a direction but not a mandate. Watch for federal AI legislation introduced in the next two quarters. If bills include preemption language, the federal-state conflict moves from theoretical to operational.
The US doesn’t yet have a unified AI governance regime. For now, compliance teams are running two playbooks simultaneously.