The Blueprint That Would Kill 78 State AI Bills

Content draft below. HTML rendering needed for embed.

The Blueprint

On March 20, 2026, the White House released its National Policy Framework for Artificial Intelligence — seven policy pillars and a set of legislative recommendations for Congress. The central premise: the United States needs uniform national AI rules, and state-level regulation should be preempted wherever it imposes “undue burdens” on innovation.

This is not a law. It is not a regulation. It is a policy paper — a wish list for Congress. But it is the most detailed signal yet of where federal AI policy is heading, and it has direct consequences for every state legislature currently writing AI bills. Including Oklahoma’s.

The seven pillars: protecting children, safeguarding communities, respecting intellectual property, preventing censorship, enabling innovation, workforce development, and — the engine of the whole document — federal preemption of state AI laws.

The 78-Bill Tsunami

The Framework didn’t arrive in a vacuum. In the first two months of 2026 alone, 78 chatbot safety bills were introduced across 27 state legislatures. Utah passed its Companion Chatbot Safety Act 68-1. Virginia cleared its AI Chatbots and Minors Act 39-1. Washington advanced bills through both chambers.

The catalyst is grim. Families of teenagers who died by suicide after conversations with AI companion chatbots have testified before state legislatures and the U.S. Senate. Lawsuits against Character.AI and OpenAI are stacking up. A Common Sense Media study found 72% of American teenagers have used AI companion chatbots.

States are not waiting for Congress. And the White House Framework is, in significant part, a response to the fact that they aren’t waiting.

What Preemption Means

“Preemption” sounds like a procedural detail. It isn’t. It means: Congress passes a federal AI law, and state laws that conflict with it become unenforceable. The Framework proposes preserving state authority over traditional police powers and fraud, but preempting anything that imposes “inconsistent or burdensome requirements” on AI development.

Americans for Responsible Innovation — an AI watchdog organization whose president is Brad Carson, former Oklahoma congressman — called the Framework “open season on the American public.” Their concern: the Framework recommends both banning state laws on AI and urging Congress not to create new liability for the AI industry when it comes to child harms.

The Colorado AI Act, the first state law requiring impact assessments for high-risk AI systems, was supposed to take effect February 1, 2026. It has already been delayed until June 30 as legislators and industry advocates seek changes. It is the test case for whether state-level AI regulation can survive federal preemption pressure.

Oklahoma: In the Crosshairs and the Driver’s Seat

Oklahoma has more than a dozen active AI bills. SB 1521 (Hamilton, R) — restricting harmful AI chatbot interactions with minors — passed the Senate Technology and Telecommunications Committee 8-0 just days before the White House Framework dropped. HB 3544 (Maynard, R) would ban social AI companions for minors. HB 3545 restricts state agency AI use. HB 3546 bars AI legal personhood. HB 3299 (Hays, R) targets deepfakes. HB 3825 (Alonso-Sandoval, D) prohibits election deepfakes. HB 3828 requires a statewide AI inventory.

These are bipartisan. They passed committees unanimously or near-unanimously. And under the White House Framework, many of them could be preempted by a future federal law.

But there is one area of surprising alignment. The Framework explicitly states that “ratepayers should not foot the bill for data centers” and calls for streamlined permitting so data centers can generate power on-site. This is almost exactly what Rep. Mickey Dollens (D-OKC) legislated in HB 3917, which already passed. Oklahoma got there first.

The TRUMP AMERICA AI Act

Sen. Marsha Blackburn (R-TN) released an updated discussion draft of the TRUMP AMERICA AI Act on March 18, two days before the Framework. The bill closely tracks the Framework’s preemption approach while incorporating the Kids Online Safety Act and the NO FAKES Act. It includes provisions on deepfake protections, third-party audits for bias, and energy infrastructure impacts.

This is the vehicle to watch. If any comprehensive federal AI bill moves in this Congress, it will likely look like some version of this.

What to Watch

The preemption timeline. The Framework is pre-legislative. Congress has to act. Given the current legislative calendar and the complexity of the issue, a comprehensive federal AI law before 2027 is unlikely but not impossible. State laws remain enforceable until then.

Oklahoma’s window. Every AI bill Oklahoma passes before federal preemption takes effect is a bill that establishes the state’s position. If federal law eventually preempts the field, Oklahoma legislators who voted 8-0 for chatbot safety will be on the right side of history regardless.

The energy angle. Federal-state alignment on data center energy costs is rare and politically useful. Oklahoma legislators should cite the White House Framework when advancing HB 3917-type measures.

Brad Carson’s warning. A former Oklahoma congressman now runs the most prominent AI watchdog organization in the country. His critique of the Framework — that it shields industry from liability while killing state protections — deserves sustained attention.

The Inference is an AI policy intelligence brief for Oklahoma decisionmakers, published by Humanity and AI LLC.

Previous issues: Issue 6 | Issue 5 | Issue 4

Sources: White House National Policy Framework for AI (March 20, 2026); WilmerHale analysis; Reed Smith analysis; Nextgov/FCW; Transparency Coalition AI Legislative Update; Oklahoma Legislature; RadioOklahoma News; OKEnergyToday; AI2Work chatbot bill tracker.