Trump's War on State AI Law.
DOJ Sues Colorado.
Senate Revolts 99–1.
On December 11, 2025, President Trump signed Executive Order 14365, directing the Department of Justice to challenge state AI laws, conditioning $42.45 billionin BEAD broadband funding on state compliance, and tasking the White House to develop a single federal “One Rulebook” for AI regulation. The order's architect was David Sacks, Trump's AI and Crypto Czar, who argued that 50 divergent state regimes would shatter the national AI industry and hand China an opening.
By April 2026, the DOJ had acted on the order's mandate: it joined Elon Musk's xAI in a federal lawsuit against Colorado Attorney General Phil Weiser (D), seeking to block Colorado SB 24-205 — the state's first-in-the-nation AI consumer protection law — before its June 30, 2026 effective date. A federal judge issued a stay. Weiser agreed to halt enforcement. The case, X.AI LLC v. Weiser (No. 1:26-cv-01515), is now the live constitutional battlefield for whether states can regulate AI at all.
Congress, however, rejected the core preemption mechanism. On July 1, 2025 — months before the Colorado lawsuit — the Senate voted 99-1 to strip a 10-year moratorium on state AI regulation from the Republican budget reconciliation bill. The lone dissenter was Sen. Thom Tillis (R-NC), in what colleagues called a possible accidental vote during a 4 a.m. vote-a-rama. The fight has since shifted from legislation to litigation and executive action.
- $42.45BBEAD broadband fundingconditioned on state AI law compliance under EO 14365 — states with AI laws deemed 'onerous' risk losing their entire broadband infrastructure allocation
- 99–1Senate vote, July 1, 2025voted to strip the 10-year state AI moratorium from the budget reconciliation bill — only Sen. Thom Tillis (R-NC) voted to keep it, in what colleagues described as a possible error
- 1:26-cv-01515X.AI LLC v. Weisercase number in the U.S. District Court for Colorado — DOJ intervened April 24, 2026; judge stayed SB 24-205 on April 28; effective date June 30, 2026 now in limbo
- 1,200+state AI bills introducedsince 2023 across all 50 states — 100+ already enacted; Sacks cited this figure as the core rationale for federal preemption
Signed December 11, 2025, EO 14365 — officially titled “Ensuring a National Policy Framework for Artificial Intelligence”— is the most aggressive federal attempt to subordinate state AI regulation in U.S. history. It operates through four levers.
1. DOJ Litigation Task Force. The order directs the Attorney General to establish a task force to identify and challenge state AI laws in federal court on constitutional grounds, including First Amendment, Commerce Clause, Due Process, and Equal Protection claims. AG Pam Bondi formally stood up the task force on January 9, 2026; it began operations January 10.
2. BEAD Funding Conditionality.The Secretary of Commerce is directed to condition eligibility for non-deployment BEAD funds — part of the $42.45 billion broadband program previously appropriated under the Infrastructure Investment and Jobs Act — on states not having AI laws deemed onerous. States that enacted consumer-protection AI mandates could lose their entire broadband allocation.
3. FTC Act Preemption Argument.The order directs DOJ and the FTC to develop a legal argument that certain state AI mandates — specifically those requiring AI systems to alter “truthful outputs” — are preempted by the FTC Act's prohibition on deceptive trade practices.
4. Legislative Framework. The White House was directed to produce, by March 11, 2026, a comprehensive evaluation of state AI laws and legislative recommendations to Congress for a single federal framework. The White House delivered that framework on March 20, 2026, calling for broad federal preemption with carve-outs for child safety, state government procurement, and data center infrastructure.
The order explicitly carves out state laws on child safety protections, AI compute and data center infrastructure, state government procurement, and “other topics” yet to be defined. That ambiguity in the carve-outs has added to the legal uncertainty: opponents argue the undefined scope makes the preemption mechanism void-for-vagueness before it even reaches court.
David Sacks, a PayPal Mafia veteran and Craft Ventures founder, served as Trump's AI and Crypto Czar from inauguration through March 26, 2026, when he exhausted his 130-day limit as a special government employee. Sacks was the intellectual architect of the “One Rulebook” framework and its primary public advocate. He posted an extended defense of the preemption position on X that became the defining statement of the administration's approach.
ONE RULEBOOK FOR AI— I wanted to share a few thoughts on AI preemption and address some of the concerns. First, this is not an “AI amnesty” or “AI moratorium.” It is an attempt to settle a question of jurisdiction. When an AI model is developed in state A, trained in state B, inferenced in state C, and delivered over the internet through national telecommunications infrastructure, that is clearly interstate commerce.
In the absence of federal preemption, 50 different states will assert their jurisdiction. Over 1,200 bills have been introduced in state legislatures with over 100 measures already passed. If every state sets its own rules, the industry could splinter into “50 different AI models for 50 different states” — a regulatory morass worse than Europe that would slow innovation, squeeze smaller startups, and leave room for China to advance.
“There must be only One Rulebook if we are going to continue to lead in AI.”
President Donald J. Trump · Truth Social · December 2025 · announcing EO 14365
U.S. leadership in AI will be DESTROYED IN ITS INFANCY if we allow 50 States, many of them bad actors, involved in RULES and the APPROVAL PROCESS. There must be only One Rulebook if we are going to continue to lead in AI. I will be signing an Executive Order very soon on this matter!
Sacks stepped down from the special government employee role on March 26, 2026 and transitioned to co-chair of the President's Council of Advisors on Science and Technology (PCAST), an advisory post with no direct policy authority. He predicted Congress would pass AI legislation within months. As of May 11, 2026, no bill has reached the Senate floor.
Colorado SB 24-205, signed by Governor Jared Polis (D) on May 17, 2024, was the first state law in the nation specifically designed to regulate “high-risk” AI systems that make or substantially affect “consequential decisions” in employment, housing, financial lending, and education. The effective date was delayed to June 30, 2026 after a failed special session in August 2025. Then xAI filed suit.
Who it covers:Developers and deployers of “high-risk” AI systems — defined as systems that are a substantial factor in making “consequential decisions” in employment, housing, credit, education, and healthcare.
Developer obligations: Use reasonable care to protect consumers from foreseeable algorithmic discrimination; disclose known risks of bias to deployers; publish a public statement summarizing risk-management practices; notify the AG and known deployers within 90 days of discovering or receiving credible reports of discriminatory risk.
Deployer obligations: Implement a risk-management policy and complete an impact assessment; conduct annual reviews for algorithmic discrimination; notify consumers in writing when a high-risk AI system makes or substantially contributes to a consequential decision affecting them; provide consumers with a right to correct data and appeal adverse decisions.
Enforcement: State Attorney General. No private right of action. Maximum civil penalties not yet specified in the bill as enacted.
Historical discrimination carve-out (contested):The law exempts differential treatment intended to “expand an applicant, customer, or participant pool to increase diversity or redress historical discrimination.” xAI and the DOJ argue this exemption constitutes an unconstitutional race-based classification under the Equal Protection Clause.
On April 9, 2026, xAI filed its complaint in the U.S. District Court for Colorado (Case No. 1:26-cv-01515), raising six constitutional claims. On April 24, 2026, the DOJ filed a complaint in intervention, joining the case on Equal Protection grounds. Notably, the DOJ's argument was not that the whole statute violates the Constitution — it was that the historical discrimination carve-out unconstitutionally compels AI systems to make race-conscious adjustments.
“We have to evaluate this law very carefully. A state-by-state patchwork of regulation poses significant challenges to the cultivation of a strong technology sector.”
Colorado Attorney General Phil Weiser (D) · on SB 24-205 · prior to the lawsuit
The same afternoon the DOJ filed its intervention, Weiser agreed — in a joint filing with xAI — to voluntarily halt enforcement of SB 24-205. The following day, April 28, 2026, the federal judge issued a formal stay. The law's June 30, 2026 effective date remains on the books; its enforcement is suspended pending a ruling on xAI's preliminary injunction motion.
Before the Colorado lawsuit, the administration's preferred tool was legislative: a 10-year moratorium on state AI enforcement, embedded in the Republican budget reconciliation bill (the “One Big Beautiful Bill Act”). The provision was authored with industry support and would have conditioned the entire $42.45 billion BEAD program on state compliance. Any state enforcing an AI consumer-protection law would have forfeited its broadband infrastructure allocation.
Duration: 10-year moratorium on state enforcement of AI-specific regulations.
BEAD conditionality: States accepting BEAD non-deployment funds would have had their AI laws automatically voided for the duration of the moratorium. States that refused to accept the condition would have lost their entire share of the $42.45B broadband program.
Effective scope:Any state AI law — consumer protection, bias mandates, transparency requirements, age verification for AI — would have been unenforceable for a decade under any interpretation.
Carve-outs: The provision as written had no carve-out for child safety laws, no exception for state government procurement, and no exception for infrastructure regulations.
The amendment to strip the moratorium was jointly offered by Senators Marsha Blackburn (R-TN), Maria Cantwell (D-WA), Ed Markey (D-MA), and Susan Collins (R-ME). The 4 a.m. vote-a-rama result: 99-1. Senator Thom Tillis (R-NC) cast the lone vote against striking the provision. Tillis later told reporters the outcome was a “silly” result and suggested his vote may have been inadvertent. After the amendment passed, the Senate voted 51-50 to pass the underlying bill, with Vice President JD Vance breaking the tie. The House passed the amended bill 218-214 on July 3. President Trump signed it into law July 4, 2025.
99 senators voted to strip the AI moratorium (i.e., voted to preserve state authority to regulate AI). This included a majority of Republicans.
1 senator voted to keep the moratorium:Sen. Thom Tillis (R-NC). Tillis called the result “silly” and said he did not know “what happened. Maybe people got tired.”
Lead amendment sponsors:Sen. Marsha Blackburn (R-TN) · Sen. Maria Cantwell (D-WA, ranking member, Senate Commerce Committee) · Sen. Ed Markey (D-MA) · Sen. Susan Collins (R-ME)
What it means: The 99-1 vote makes future legislative preemption of state AI laws functionally impossible in this Congress without substantial modification. The administration pivoted exclusively to litigation and executive action after July 1, 2025.
The BEAD conditionality mechanism survives in Executive Order 14365 even after the legislative moratorium was stripped. The EO directs the Secretary of Commerce to condition BEAD non-deployment funds — money appropriated under the 2021 Infrastructure Investment and Jobs Act for broadband infrastructure in underserved communities — on state AI law compliance. States with AI laws that the Commerce Department deems “onerous” could lose broadband funding for rural, tribal, and low-income communities.
The coercion problem. In NFIB v. Sebelius(2012), the Supreme Court held that the federal government cannot coercively withhold previously appropriated funds to compel state policy changes. The BEAD program was appropriated in 2021 under the Biden administration for a specific purpose: rural and underserved broadband infrastructure. Conditioning it on AI law compliance — an entirely different policy domain — is arguably the same constitutional violation the Court identified in the ACA's Medicaid expansion coercion.
The Commerce Department's evaluation. EO 14365 required Commerce to publish, by March 11, 2026, an evaluation of state AI laws deemed onerous. As of late April 2026, no such evaluation has been publicly released. Without a published list of targeted laws, no state can know whether its AI legislation puts its BEAD allocation at risk.
State AG response. At least 36 states have formally objected to the BEAD conditionality mechanism via letters to Commerce. Multiple state attorneys general have indicated they will seek injunctive relief if the Commerce Department moves to withhold funds.
Colorado is not alone. The federal preemption effort targets a sprawling and rapidly expanding body of state AI law. Sacks cited 1,200+ state AI bills introduced nationwide since 2023, with 100+ enacted. The patchwork is real.
Colorado SB 24-205(effective June 30, 2026, now stayed) — the first state “high-risk AI” bias law; now the subject of the federal lawsuit. Governor Polis (D) and AG Weiser (D) have both expressed reservations about the law's design.
California — multiple laws effective January 1, 2026 — covering generative AI disclosure, frontier model safety, AI chatbot identification, healthcare communications, and algorithmic pricing. California SB 942 major platform transparency obligations take effect August 2, 2026.
Texas Responsible AI Governance Act (RAIGA) — effective 2026; emphasizes enterprise AI transparency, documentation requirements, and red-teaming obligations for high-risk systems.
New York RAISE Act — chapter amendment signed March 27, 2026; effective January 1, 2027; DFS oversight; AG enforcement up to $1M per violation / $3M per pattern.
36 states have formally moved to block federal preemption of their AI laws as of March 2026, setting up a constitutional fight the federal courts have not yet resolved.
On March 20, 2026, the White House released its National Policy Framework for Artificial Intelligence, the legislative recommendations required under EO 14365. The document outlines six objectives and asks Congress to enact them into law.
1. Protect childrenfrom online harms and sexual exploitation — explicit carve-out from preemption for state child safety laws.
2. Safeguard communitiesby barring ratepayers from subsidizing AI data centers and streamlining data center permitting — carve-out for infrastructure regulation.
3. Respect intellectual property and support fair use of training data for AI systems.
4. Prevent censorship by prohibiting AI systems from filtering lawful political expression in a manner inconsistent with the First Amendment.
5. Enable innovationby removing regulatory barriers to AI development, expanding access to federal testing environments, and pre-empting state AI laws that impose “undue burdens.”
6. Educate Americans through AI workforce development, apprenticeships, and retraining programs.
Congressional prospects for the framework remain dim. The 99-1 Senate vote demonstrated that even Republican senators with strong pro-deregulation track records are unwilling to strip state AI authority wholesale. The White House framework's broad preemption language — “preventing states from regulating use of AI for activity that would be lawful if performed without AI” — has been criticized by state AGs of both parties as overreach that would nullify consumer protection laws predating AI entirely.
Major AI and technology companies were the primary lobby behind the legislative moratorium. Google, Meta, Microsoft, Apple, OpenAI, and xAI all backed federal preemption of state AI law, either directly or through trade associations including TechNet, the Software Alliance, and the Information Technology Industry Council. The argument was Sacks's: 50 incompatible state regimes make national-scale AI deployment operationally impossible and competitively dangerous.
What industry wanted: Federal law preempting state AI regulation for at least 10 years, eliminating the compliance cost of state-by-state rules on training data, model disclosure, bias audits, consumer rights, and high-risk system requirements.
What industry got from Congress: A 99-1 rejection of the moratorium. No federal AI legislation enacted as of May 11, 2026.
What industry got from the executive branch: DOJ litigation challenging the most aggressive state laws (Colorado first); BEAD conditionality threat (legally uncertain, unenforced); White House legislative framework (Congress not yet moved).
The litigation bet:xAI's Colorado lawsuit, backed by the DOJ, is now the highest-stakes venue for defining whether state AI consumer protection laws survive constitutional scrutiny. A win in Colorado would create precedent the DOJ could use to challenge laws in Texas, New York, and California.
“If America wants to win the AI race, a confusing patchwork of regulation won't work. Fifty different state AI models would be worse than Europe.”
David Sacks · AI & Crypto Czar · January 2026 · on the preemption rationale
As of May 11, 2026, the federal-state AI fight is active on three simultaneous fronts.
Front 1: X.AI LLC v. Weiser, D. Colo.The preliminary injunction motion is the most time-sensitive. If the court grants the injunction, SB 24-205 is blocked past its June 30 effective date. If it denies, Colorado's law goes live and the DOJ faces immediate appeals with national precedent at stake. A ruling is expected before June 30.
Front 2: BEAD conditionality. The Commerce Department's still-unpublished evaluation of state AI laws is overdue since March 11. If Commerce releases the list and begins withholding BEAD funds, 36+ state AGs are poised to file for injunctive relief under NFIB v. Sebelius. This front has not yet produced a court case.
Front 3: Congressional legislation.The White House framework calls for federal AI legislation with preemption authority. Given the 99-1 Senate vote, passage of broad preemption language is unlikely in the current Congress. Narrower legislation — federal baseline standards for high-risk AI with partial field preemption — remains a theoretical possibility but has no Senate sponsor as of May 2026.
Trump signed EO 14365 on December 11, 2025, threatening $42 billionin broadband funding and launching a DOJ litigation machine against state AI laws. The Senate rejected the legislative version 99-1 in July 2025. The DOJ joined xAI's lawsuit against Colorado's AI law in April 2026; a federal judge stayed the law days later. The constitutional question — whether states can mandate AI bias audits, consumer disclosures, and impact assessments at all — will be decided in a federal courthouse in Denver before June 30, 2026. Every other state AI law in America is watching.
Tier 1: White House executive orders, DOJ press releases and court filings, Colorado General Assembly legislative text, Senate Commerce Committee press releases. Tier 2: Bloomberg Law, Paul Hastings, Jenner & Block, Ropes & Gray for legal analysis. Tier 3: Rocky Mountain Voice, Colorado Politics, Time, NPR, CNBC for news coverage. David Sacks posts cited directly from X post IDs. Where legal analysis differs between sources, the primary court record (DOJ complaint, PACER docket) controls.