The White House released its National Policy Framework for Artificial Intelligence on March 20, calling for Congress to pass federal legislation that would preempt the patchwork of state AI laws now emerging across the country.
Within a week, Washington Governor Bob Ferguson signed two AI bills into law. Oregon’s chatbot bill awaits Governor Kotek’s signature after passing both chambers unanimously. States aren’t waiting for Congress.
White House Framework: Preempt the States
The four-page framework outlines the administration’s vision for AI regulation. The core message: create one national standard and stop states from building “fifty discordant” regulatory regimes.
Key recommendations:
The framework calls for federal preemption of state AI laws that “impose undue burdens,” while preserving states’ ability to enforce consumer protection, fraud prevention, and child safety laws. Specifically, it wants Congress to prevent states from regulating AI model development or holding AI developers liable for what third parties do with their systems.
On child protection, the framework recommends age-assurance requirements for AI platforms—though it explicitly says parental attestation would satisfy this, rather than age verification. It supports the TAKE IT DOWN Act prohibiting nonconsensual intimate imagery and deepfakes.
The framework also pushes back against what it calls government “coercion” of tech companies to moderate content based on “partisan or ideological agendas.” Critics note the irony given the administration’s own “Unbiased AI Principles” requirements for federal AI procurement.
What it doesn’t do:
The framework isn’t law. It’s a set of recommendations for Congress. And Congress has repeatedly rejected federal AI preemption language—declining to include it in both the One Big Beautiful Bill Act and the National Defense Authorization Act. GOP House leaders pledged support, but Democrats introduced competing legislation. The framework’s future is uncertain.
Washington Signs Two AI Bills
On March 25, Governor Bob Ferguson signed two AI bills into law—five days after the White House released its preemption framework.
House Bill 2225 (Chatbot Regulation)
Washington becomes the second state to pass legislation specifically targeting AI companions. The law requires:
- Chatbots must disclose they’re not human at the start of every conversation
- Ongoing conversations require disclosure every three hours
- Chatbots cannot claim to be human
- Operators must implement suicide prevention protocols
- Crisis intervention resources must be provided when needed
The law takes effect January 1, 2027, and includes a private right of action—users can sue operators who violate it.
House Bill 1170 (AI Content Provenance)
Ferguson personally asked legislators to craft this bill targeting AI-generated misinformation. When content is substantially modified using generative AI, that information must be traceable through watermarks or metadata.
This law takes effect January 1, 2028, giving platforms more time to implement technical requirements.
Oregon Chatbot Bill Awaits Signature
Oregon’s Senate Bill 1546 passed the Senate 26-1 and House 52-0 on March 5. Governor Tina Kotek has until the end of this week to sign or veto.
The bill is arguably the most detailed chatbot regulation in the country:
For all users:
- Disclose that users are talking to AI, not a human
- Prevent outputs that could cause suicidal ideation
- Report incidents to Oregon Health Authority
Additional requirements for minors:
- Initial notification that the chatbot may be unsuitable for children
- Hourly reminders that they’re speaking with AI
- Prohibition on deceptive identity claims
- Hourly break reminders
- No sexually explicit content
- No addictive engagement algorithms
- No emotionally manipulative messages when users try to disconnect
The law creates a private right of action—anyone who suffers “ascertainable harm” can sue for damages and injunctive relief. That’s significant: most state AI laws rely on attorney general enforcement. Private lawsuits create a different kind of accountability.
Utah Passes Nine AI Bills
Governor Spencer Cox signed multiple AI bills during the 2026 legislative session, including HB 276, the Digital Voyeurism Prevention Act.
The bill requires AI operators to embed provenance data that lets users determine if an image was created or altered using AI. It also enacts deepfake protections.
Utah joins the growing list of states requiring AI content transparency, a trend the White House framework explicitly supports even while opposing other state-level regulation.
Vermont’s AI Election Law Now Active
Vermont’s S.23 went into effect after Governor Phil Scott signed it on March 5.
The law requires any campaign media within 90 days of an election that includes AI-generated images, audio, or video to display a disclosure. The disclosure must be “visible to the average viewer and large enough to read easily.”
The law also prohibits creating materially false AI-generated information targeting voters and gives the attorney general enforcement authority under consumer protection law. Satire and parody are exempted.
With midterms approaching, Vermont is one of the first states with active AI election rules. More will follow.
What’s Still Moving
Bills signed or awaiting signature:
- Oregon SB 1546 (chatbot safety) - awaiting governor
- Washington HB 2225 (chatbot disclosure) - signed
- Washington HB 1170 (AI provenance) - signed
- Utah HB 276 (deepfake/provenance) - signed
Bills advancing:
- Idaho S 1297 (chatbot safety) - passed Senate, House second reading
- Pennsylvania SB 1090 (SAFECHAT Act) - passed Senate 49-1, in House committee
- Arizona HB 2311 - advancing before session end
- Georgia SB 540 - advancing before April 6 session end
The numbers: Six weeks into the 2026 legislative season, 78 chatbot bills are alive in 27 states. The wave shows no signs of slowing.
The Preemption Standoff
The White House framework creates an interesting dynamic. The administration wants Congress to preempt state AI laws. But Congress hasn’t acted, and states aren’t waiting.
Washington and Oregon passed their chatbot laws knowing the White House position. Pennsylvania’s SAFECHAT Act cleared the Senate 49-1 the same week the framework was released. States are betting that either Congress won’t act, or that their laws fall within the framework’s carve-outs for consumer protection and child safety.
The framework does preserve states’ traditional police powers. But the language is vague enough that courts would ultimately decide what that means. A state chatbot law requiring specific disclosures might survive; a state law imposing liability on AI developers might not.
For companies, the immediate calculus remains the same: comply with state laws as they pass, because federal preemption isn’t here yet and might never arrive. Washington, Oregon, Pennsylvania (if passed), and Idaho (if passed) all target January 2027. That’s nine months away.
For everyone watching this space, the pattern is clear. The federal government publishes frameworks. States pass laws. The gap between policy recommendations and actual regulation keeps widening.