← back to home

What we’re asking for, and why

Who decides?

In 30 seconds

Right now, nobody with legal power checks the most powerful AI before it’s released. Congress hasn’t passed a law to oversee it, so a few companies decide alone what’s safe. [5]

So we’re asking for rules at both levels:

  • Your state free to protect you now, and to keep stronger protections even after a national law passes.
  • A strong national law from Congress, too: independent safety testing before release, and a watchdog that can stop AI that isn’t safe.

This isn’t states instead of Congress, and it isn’t “state rules are better.” It’s both, because of where we are: no national law yet, and real harm already happening.

Both parties are on this. 99 senators from both parties voted to keep states able to act. [24] Attorneys general from 36 states and territories, red and blue, agreed. [116] State AI safety laws have been signed by Democratic and Republican governors. [94] [103]

Your move

Ask your state to keep protecting you, and ask Congress for a strong national law too.

Contact your reps → See your candidates →

Why both levels

Your state, right now

What you can win this year.

  • They exist today. State laws are the only AI safety laws on the books in the U.S. right now. [5]
  • States move faster. They’ve already acted on deepfake nudes, AI chatbots and kids, and telling people when they’re talking to AI. Attorneys general from both parties say blocking states would weaken their ability to respond quickly. [116]
  • They test what works. When a state rule doesn’t work, it can be fixed fast. Colorado replaced its own AI law in 2026 after learning from the first version. [39]
  • They’re closer to you. Your state attorney general can take a company to court for hurting people in your state. [116]

Congress, too

Some things only a national law can do.

  • AI doesn’t stop at state lines. The most powerful systems are used everywhere at once.
  • Only Congress can create a national watchdog with the power to test powerful AI before release and stop it if it isn’t safe. No state can do that for the whole country.
  • One clear standard helps everyone plan, including small companies. Unpredictable rules hurt them most. [61]

Keeping states able to act, from both parties

99 U.S. Senators from both parties

Voted 99 to 1 in July 2025 to remove a 10 year ban on state AI laws. The fix was led by Sen. Marsha Blackburn (R, Tennessee) and Sen. Maria Cantwell (D, Washington). [24]

36 state and territory attorneys general

From both parties, including Arizona, California, Idaho, Kansas, Louisiana, Ohio, South Carolina, Tennessee and Utah. They wrote to Congress in November 2025 opposing a federal ban on state AI laws. [116]

Republican governors

Gov. Sarah Huckabee Sanders (Arkansas), Gov. Spencer Cox (Utah) and Gov. Ron DeSantis (Florida) criticized the push to block state AI rules. [117]

Anthropic and Encode

Anthropic, an AI company, endorsed California’s AI safety law, SB 53. Encode, a youth led advocacy group, sponsored it. [100] [118]

Pushing for Washington only

Open +

Sen. Ted Cruz (R, Texas)

Wrote the 10 year ban on state AI laws that the Senate removed in 2025. [24]

Sam Altman, Palmer Luckey and Marc Andreessen

Tech leaders from OpenAI, the defense company Anduril and the investment firm Andreessen Horowitz backed that 10 year ban. [24]

OpenAI

Asked the White House in March 2025 to override state AI laws, in exchange for letting the government test its models. [119]

The White House

A December 2025 order told the Justice Department to challenge state AI laws and called for one national standard. [2]

Leading the Future

An AI industry super PAC network funded by people like OpenAI president Greg Brockman and Andreessen Horowitz. It wants to get rid of the “patchwork” of state rules. [120] [96]

Microsoft and Andreessen Horowitz

In 2024 they jointly opposed new AI rules, arguing they would burden startups. [101]

Not everyone here wants zero rules. A bipartisan House draft from Rep. Jay Obernolte (R, California) and Rep. Lori Trahan (D, Massachusetts) would add national AI safety rules but pause some state laws for three years. [121] And Sen. Blackburn, who defended states in 2025, now has her own draft for one federal rulebook with strong protections for kids. [122]

The Washington only side’s best argument

Open +

Fifty different rulebooks could be confusing and expensive, especially for small companies without big legal teams. AI doesn’t stop at state lines. Supporters say one clear national standard would help the U.S. compete and stay secure. [119] [102] Research does find that unpredictable rules hurt small companies the most. [61]

The real risks, both ways

Open +

If every state does its own thing

  • Rules can clash, and following many of them costs money, which hits small companies hardest. [61]
  • AI companies say a patchwork of different state rules slows them down. [119]

If states get overridden before Congress acts

  • There would be no AI safety rules anywhere until Congress passes some, and it hasn’t yet. [121] [5]
  • States would lose the power to respond quickly to new harms, like they did with deepfakes. [116]
  • The fight itself creates uncertainty, which researchers say also hurts innovation. [61]

Where this site lands

Both. We want a strong national law that sets a real floor: independent testing, a watchdog that can stop unsafe AI, and protections for kids and workers. Until Congress passes one, blocking states just means no protections at all. And after it passes, states should still be able to go further. A national floor, not a ceiling.

See something wrong or missing? Email hey@commonlifeproject.com and I’ll fix it.