Ninety-nine senators agreed on something in July 20251. A decade-long freeze on state AI oversight went too far. Their 99-1 vote against the moratorium set the terms for a fight that ran all through 2026. Washington kept pushing preemption anyway. A December 2025 executive order, a March 2026 legislative framework, and two rival super PACs turned one defeat into a standing campaign3. OpenAI's Chris Lehane, Anthropic's Jack Clark and a widening circle of policy chiefs steered their employers toward opposite positions.
A Moratorium Meets the Senate
House Republicans inserted a 10-year ban on state AI regulation into a tax bill in May 20251. Their bet was simple: one federal standard beats fifty competing ones. The Senate disagreed two months later, almost to a member, stripping the provision 99-1 and leaving states free to legislate on their own schedules. That vote followed President Trump's January 2025 executive order directing agencies to draft a national AI Action Plan2. Published that summer, the plan set three aims: faster innovation, more AI buildout, and American leadership in AI diplomacy and security.
The Senate voted 99 to 1 in July 2025 to strip a 10-year ban on state AI laws
Senate votes on stripping the state AI law moratorium from the tax bill, one cube per senator
Each cube is one senator's vote: the 99 magenta cubes voted to strip the moratorium, and the single gray cube voted to keep it.
Source: Wikipedia [1]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Value |
|---|---|
| Voted to strip the moratorium | 99 |
| Voted to keep the moratorium | 1 |
Congress handed the White House a loss on procedure. The administration's preference for a single federal rulebook survived the vote intact. Momentum moved from bills to executive orders, and that shift defined the next phase of the fight.
Executive Orders Try Again
Trump signed Executive Order 14365 in December 20251. Its title named the goal: a national policy framework for AI. The order sought to preempt state rules through executive power, after Congress declined to do the job by statute. A second push came in March 2026, when the White House handed Congress model language for another legislative run.
Two attempts inside four months read as persistence rather than confidence. An executive order skips the votes. It also carries a risk a statute avoids. Courts, a future administration, or simple defiance from state attorneys general can undo an order far faster than they can undo a law.
States Write Their Own Rules
California moved first among the large states1. Gov. Gavin Newsom signed the Transparency in Frontier Artificial Intelligence Act, known as SB 53, in September 2025. The law took effect Jan. 1, 2026. It made California the first state with a statute aimed squarely at frontier model development, requiring safety-test disclosure and shielding whistleblowers who report a breach. Texas chose a different template. Gov. Greg Abbott signed the Texas Responsible AI Governance Act on June 22, 2025, creating a state AI council alongside developer and deployer duties that took effect the same January.
New York took the slowest path1. Gov. Kathy Hochul signed the RAISE Act on Dec. 19, 2025, adding transparency and safety duties for frontier developers, with an effective date pushed to Jan. 1, 2027, well past its peer states. Colorado went the other direction. Its broad AI Act, first due Feb. 1, 2026, slipped to an expected June start. Then Gov. Jared Polis signed SB 26-189 on May 14, 2026, repealing the original framework and replacing it with lighter requirements. Four states, four calculations about how much regulation their AI sectors could carry.
Where the Money Went
Capital followed conviction. Silicon Valley investors and firms pledged up to $200 million in August 2025 to two pro-AI super PACs, Meta California and Leading the Future1. Both were built to reward candidates who back preemption. Leading the Future registered Aug. 15, 2025 in Henderson, Nevada, and grew into the larger vehicle3. Andreessen Horowitz co-founders Marc Andreessen and Ben Horowitz committed $50 million in 2026. OpenAI President Greg Brockman and Anna Brockman added $25 million. Palantir co-founder Joe Lonsdale joined the donor roll. Fundraising passed $140 million under co-leaders Zac Moffatt and Josh Vlasto. Lehane advised on the PAC's formation and on Brockman's political spending directly, and Andreessen Horowitz government-affairs head Collin McCune took part in pre-launch planning.
Anthropic went the other way. The company donated $20 million in February 2026 to Public First Action, a group backing candidates open to AI rules1. Read that as a wager that guardrails serve the company's long-term interest more than a deregulated market would. Two labs, drawing from similar capital pools, bet on opposite theories of what protects their position.
Greg and Anna Brockman gave Leading the Future $25 million, and Anthropic gave the other side $20 million
Named donations on each side of the preemption fight, in millions of dollars
Each bar is one donor's commitment: blue bars went to Leading the Future, which backs preemption, and the magenta bar is Anthropic's gift to Public First Action, which backs candidates open to AI rules.
Sources: Wikipedia [1]; Wikipedia [3]. Chart by Ryan Elliott Dennis.
The numbers behind this chart
| Item | Value | Note |
|---|---|---|
| Marc Andreessen and Ben Horowitz | $50 million | Committed in 2026 |
| Greg and Anna Brockman | $25 million | Greg Brockman is OpenAI's president |
| Anthropic | $20 million | Donated in February 2026 |
The Policy Chiefs Behind the Fight
Names explain the split as well as dollars do. Lehane runs global affairs for OpenAI and shaped Leading the Future from its earliest planning conversations, which gives that push an operator who ran Democratic campaigns for years3. Clark co-founded Anthropic and leads its policy team5. He now works alongside Mariano-Florentino Cuéllar, the company's first chief global affairs officer, appointed Aug. 4, 2026 during separate friction with the Pentagon over a supply-chain risk label and tighter export controls. Kent Walker holds the equivalent seat at Google, Brad Smith at Microsoft, and Joel Kaplan at Meta. Those three employers signed differently across the coalitions of 2026.
Microsoft's clearest move of the year landed in July6. The company hosted and signed the "Open Weights and American AI Leadership" letter on July 24, 2026, alongside 25 founding signatories including Nvidia, Meta, IBM, Dell and Palantir4. More than 270 companies and organizations had added their names by Aug. 3. Its case ran in plain terms. "Open weight models—AI models that anyone can download, inspect, modify, and run on their own infrastructure—are an important part of that foundation because they make advanced AI more accessible, adaptable, and widely available."
The Microsoft-hosted letter saved its sharpest claim for the safety question. "In fact, openness may be one of the most important paths to AI safety and security."6 One sentence carried the reasoning: "Just as open-source software demonstrated that transparency can be more secure than obscurity, AI safety may depend on giving more people the ability to test and strengthen the models on which society relies." Look at what that does to the usual alignment of arguments. Firms asking for fewer state rules grounded the ask in safety, the same value those state laws claim.
Every one of these executives answers to a board weighing legal risk against product speed, and 2026 gave each a reason to bet differently. Lehane's employer chases consumer scale, which a fifty-state rulebook complicates. Clark's employer sells safety as a feature, a pitch that a credible regulatory floor reinforces. The gap between those two calculations explains why Washington's preemption fight keeps going, better than any single vote or order does.
Cruz chairs the Senate Commerce Committee that wrote the moratorium, and he kept building the case for a single federal standard after the Senate stripped it. At a Sept. 24, 2025 hearing he titled "Will AI's Future Be American?," he turned to the states already writing AI statutes of their own. "Existing criminal, tort, contract, child safety, and consumer protection laws already govern AI, making state efforts to impose AI-specific regulations—such as those in Colorado, California, or in more than a thousand other state-level bills—needlessly duplicative and punitive," Cruz said7.
Blackburn voted to strip that same moratorium, and she released her own case for keeping the states in charge the night of the vote. The Tennessee Republican argued that Congress had yet to pass the federal privacy and child-safety bills the moratorium's backers pointed to as a substitute. "Until Congress passes federally preemptive legislation like the Kids Online Safety Act and an online privacy framework, we can't block states from standing in the gap to protect vulnerable Americans from harm, including Tennessee creators and precious children," Blackburn said on July 1, 20258.
By the numbers
- Ninety-nine to one: the Senate vote stripping a proposed 10-year moratorium on state AI laws in July 20251.
- Two hundred million dollars: the combined Silicon Valley pledge to Meta California and Leading the Future, announced August 20251.
- Leading the Future raised more than $140 million by 2026, per its own FEC filings3.
- Anthropic donated $20 million to Public First Action in February 2026, backing regulation-friendly candidates1.
- More than 270 companies and organizations had signed the "Open Weights and American AI Leadership" letter by Aug. 3, 20266.
- Jan. 1, 2027: New York's RAISE Act effective date, the latest among the four state laws examined here1.
What to watch
Colorado's rewritten SB 26-189 gives other states a template while Congress keeps failing to preempt: step back toward lighter duties rather than repeal1. Watch whether New York's RAISE Act reaches its January 2027 start date unchanged, given Colorado's reversal earlier in 2026. Leading the Future's spending through the midterm cycle answers a sharper question3. Does $140 million buy seats? That is the real test of whether money does what two orders and a Senate vote left undone. Anthropic's parallel bet through Public First Action runs the same experiment from the opposite direction.
Sources
- Wikipedia contributors, "Regulation of Artificial Intelligence in the United States," Wikipedia, retrieved Sept. 4, 2026, https://en.wikipedia.org/wiki/Regulation_of_artificial_intelligence_in_the_United_States
- The White House, "AI Action Plan," AI.gov, July 23, 2025, https://www.ai.gov/action-plan
- Wikipedia contributors, "Leading the Future," Wikipedia, retrieved Sept. 4, 2026, https://en.wikipedia.org/wiki/Leading_the_Future
- Tobi Opeyemi Amure, "Microsoft Just Took Sides in AI Policy Fight," TheStreet, July 26, 2026, https://www.thestreet.com/technology/ai/microsoft-just-took-sides-in-ai-policy-fight
- CNBC Staff, "Anthropic Names Global Affairs Chief as Trump Tensions Persist," CNBC, Aug. 4, 2026, https://www.cnbc.com/2026/08/04/anthropic-names-global-affairs-chief-as-trump-tensions-persist.html
- Open weights coalition signatories, "Open Weights and American AI Leadership," Microsoft, July 24, 2026, https://www.microsoft.com/en-us/corporate-responsibility/topics/open-weight/
- U.S. Senate Committee on Commerce, Science, & Transportation, "ICYMI: 'Will AI's Future Be American?'," Sept. 24, 2025, https://www.commerce.senate.gov/press/rep/release/icymi-will-ais-future-be-american/
- Marsha Blackburn, "Blackburn Leads Effort to Remove AI Moratorium from Big Beautiful Bill, Protecting Tennessee Creators & Kids from Harm," Office of Senator Marsha Blackburn, July 1, 2025, https://www.blackburn.senate.gov/2025/7/technology/video-blackburn-leads-effort-to-remove-ai-moratorium-from-big-beautiful-bill-protecting-tennessee-creators-kids-from-harm