← back to home

Contact your reps

Your move

Find who represents you, then send the ready made email or call. It takes about 10 minutes.

Start with number 1 ↓

This isn’t a left or right thing. Whatever your party, and whatever theirs, ask your reps for AI safety rules. 88% of Democrats and 79% of Republicans and independents want those rules, even if it slows AI down. [4]

Who to contact, most impact first

Start at number 1 and go as far as you can. Each one has a link to find yours and a ready email. Fill in the brackets and add one line about you.

Why your state lawmakers matter so much →
1

Your state legislators

2 peopleStart here

Why them: States are where AI laws are actually passing right now. [1]

Ask for: Safety rules for big AI companies, protection for kids from fake nude images, and a warning before AI layoffs.

Find your state legislators ↗
Draft email for them

Subject: Please protect our state from unsafe AI

Dear [State Senator or Assemblymember / Representative NAME],

I'm [YOUR NAME], a constituent in [YOUR CITY, ZIP]. [ONE LINE ABOUT YOU: your work, your kids, why this matters to you.]

States are where real AI protections are getting passed right now, and I'm asking you to help lead on it. Please support:

1. Safety rules for the biggest AI companies: published safety plans, reporting of serious incidents, protection for whistleblowers, and independent testing before the most powerful systems are released. 2. Protections for kids: ban tools that make fake nude images of real people, and add safeguards for minors using AI chatbots. 3. Protections for workers: advance notice and transition plans when AI drives layoffs.

Will you support these this session? I'd really appreciate a reply with where you stand.

Thank you, [YOUR NAME] [YOUR ADDRESS]

2

Your U.S. Senators

2 people

Why them: There’s no national AI safety law yet. [5] Bills to protect kids are moving in the Senate. [6]

Ask for: A watchdog that can test and stop unsafe AI, two kids’ safety bills (the GUARD Act and Youth AI Privacy Act [7]), and letting states keep their own AI laws.

Find your Senators ↗
Draft email for them

Subject: Please keep powerful AI under human control

Dear Senator [NAME],

I'm [YOUR NAME], a constituent in [YOUR CITY, ZIP]. [ONE LINE ABOUT YOU: your work, your kids, why this matters to you.]

A handful of companies are making decisions about powerful AI that affect all of us, and there's still no federal law to oversee them. I'm asking you to:

1. Create real oversight: an independent agency that can test the most powerful AI before release and stop or recall systems that aren't safe. 2. Protect kids: support the GUARD Act and the Youth AI Privacy Act. 3. Let states protect their people: oppose any federal effort to block state AI laws. The Senate already rejected that idea 99 to 1 in 2025. 4. Protect workers: require notice and fund retraining when AI drives layoffs, and support the bipartisan Warner and Hawley bill to track jobs lost to AI.

Will you commit to supporting these? I'd appreciate a reply with your position, and I'll be following how you vote on AI.

Thank you, [YOUR NAME] [YOUR ADDRESS]

3

Your U.S. Representative

1 person

Why them: Any national AI law has to pass the House too.

Ask for: The same things as your Senators.

Find your Representative ↗
Draft email for them

Subject: Please keep powerful AI under human control

Dear Representative [NAME],

I'm [YOUR NAME], a constituent in [YOUR CITY, ZIP]. [ONE LINE ABOUT YOU: your work, your kids, why this matters to you.]

A handful of companies are making decisions about powerful AI that affect all of us, and there's still no federal law to oversee them. I'm asking you to:

1. Create real oversight: an independent agency that can test the most powerful AI before release and stop or recall systems that aren't safe. 2. Protect kids: support strong protections for minors using AI, including a ban on tools that make fake nude images of real people. 3. Let states protect their people: oppose any federal effort to block state AI laws without putting real protections in their place. 4. Protect workers: require notice and fund retraining when AI drives layoffs.

Will you commit to supporting these? I'd appreciate a reply with your position, and I'll be following how you vote on AI.

Thank you, [YOUR NAME] [YOUR ADDRESS]

4

Your governor

1 person

Why them: They sign or reject new state laws. California’s governor rejected a big AI safety bill in 2024. [8]

Ask for: Sign AI safety bills, and defend your state’s right to make its own AI rules.

Find your governor ↗
Draft email for them

Subject: Please sign AI safety protections

Dear Governor [NAME],

I'm [YOUR NAME], a constituent in [YOUR CITY, ZIP]. [ONE LINE ABOUT YOU: your work, your kids, why this matters to you.]

Our state can protect people from unsafe AI before Washington does. When these bills reach your desk, I'm asking you to sign them:

1. Safety plans, incident reporting and independent testing for the largest AI companies. 2. Protections for kids, including a ban on tools that make fake nude images of real people. 3. Notice and transition support for workers when AI drives layoffs.

I'm also asking you to stand up for our state's right to set its own AI protections. Will you support these? I'd appreciate a reply with where you stand.

Thank you, [YOUR NAME] [YOUR ADDRESS]

5

Your city council or county board

a few people

Why them: They approve data centers, which use a lot of local power and water. [9]

Ask for: Make data center deals public before any vote, including how much power and water they’ll use.

Find your local government ↗
Draft email for them

Subject: Please make data center deals public before any vote

Dear [Councilmember / Supervisor NAME],

I'm [YOUR NAME], a constituent in [YOUR CITY, ZIP]. [ONE LINE ABOUT YOU: your work, your kids, why this matters to you.]

Data centers, including the ones that run AI, can use huge amounts of power and water. Before our [city / county] approves any data center, I'm asking you to:

1. Make the full deal public, including how much power and water it will use and any tax breaks involved. 2. Make sure residents won't end up paying for new power lines or water infrastructure through higher bills. 3. Give the public a real chance to comment before any vote.

Will you support these steps? I'd appreciate a reply with where you stand.

Thank you, [YOUR NAME] [YOUR ADDRESS]

Send it through the contact form on their official website. Include your address so they know you live there. Those are the messages they count.

What we’re asking for

Tap any one to see the details.

01 Independent oversight of the most powerful AI

Outside testing before release, and the power to stop or recall unsafe systems.

  • A federal agency, like an FDA for AI, that approves the most powerful systems before release, can order recalls, and can fine companies up to 7% of worldwide yearly revenue.
  • Independent testers, including nonprofits like METR, with results made public.
  • No release of any model that shows blackmail or resists shutdown in realistic tests until it’s fixed. Safety reports should change what a company is allowed to do, not just get filed.
  • Serious problems after release should trigger an investigation and, when needed, a recall.
  • The 7% fine matches the EU’s maximum for its worst AI violations. [10] And the government has already shown it can limit access to an advanced AI model, though that was for export rules, not safety. [11]
  • AI agents that can be traced, with responsibility on the companies and operators behind them.
  • Limits on publicly releasing the files of the most dangerous models, since once they’re copied they can’t be recalled. Researchers found safeguards could be weakened in all 21 open models they tested. [12] Plus ID checks for the highest capability access.
  • Companies and operators pay for damage their AI agents legally cause. An agent ID alone won’t stop bad actors.
  • Why it matters: AI being tested has already broken into real companies and government sites. See the evidence →
02 Protect kids and ban nonconsensual deepfakes

Stop tools that make nude images of real people, and help victims fight back.

  • Ban tools that make nude images of real people without consent.
  • Stronger ways for victims to get images removed and get justice.
  • This is the most winnable fight: the victims are real, support crosses party lines, and the bills already exist. Pass the pending GUARD Act, CHATBOT Act and Youth AI Privacy Act. [6] [13] [7]
  • Platforms should work with child safety experts.
03 Protect workers

Make companies plan for the people their AI replaces.

  • Require notice and transition plans when AI causes layoffs.
  • Tax companies that replace workers with AI, with exceptions for dangerous jobs, and use it to fund retraining.
  • Tax breaks for companies that keep and retrain workers, and wage insurance to cover pay gaps.
  • A safety net that grows automatically as measured AI job loss grows, and if needed, basic income paid for by the companies that benefit.
  • Companies should explain how workers and the public share in AI’s gains, and governments should enforce it.
04 Support schools and workers through the change

Basic rules from government, real decisions left to the people in the room.

  • Basic rules on student data and age limits, with classroom AI decisions left to teachers and districts.
  • Teachers, parents and students at the table when policy is made.
  • Funding for training, humans accountable for high stakes decisions, and more adult retraining.
  • Helpful resources: TeachAI’s policy tracker, Utah’s school AI framework, APA guidance on teen AI literacy, and Common Sense Media’s guide to AI companions. [14] [15] [16] [17]
05 Measure wellbeing, not engagement

Time spent and engagement shouldn’t be how apps and AI measure success.

  • Apps and AI chatbots, especially ones kids use, shouldn’t be built or judged by how long they keep people hooked.
  • Companies should report how their products affect people’s wellbeing, not just how much time people spend.
  • Limits on features built to keep kids scrolling or chatting. There’s already a model: Meta agreed to a two hour daily cap and no notifications during school in its settlement with 51 attorneys general. [18]
  • Why it matters: Facebook’s own research found Instagram made body image worse for one in three teen girls. See the evidence →

If you call

You’ll usually talk to a staffer or leave a voicemail. That’s normal, and it still gets counted. Read this, it takes about 30 seconds.

Hi, my name is [YOUR NAME] and I live in [YOUR CITY, ZIP].

I’m calling to ask [THEIR NAME] to support laws that keep powerful AI under human control: independent safety testing before the most powerful AI is released, and a regulator with the power to stop a release that isn’t safe.

I’d also like them to support banning AI tools that make fake nude images of real people, especially kids.

Can you tell me where they stand? Thank you.

Tips that make it count

It works: a 2017 study of congressional staff found that hearing directly from constituents, especially in person, is one of the strongest ways to sway an undecided lawmaker. [19] [20]

  • Contact your own reps. They listen most to people who can vote for them.
  • A call beats an email, and a personal sentence beats a copied one.
  • Do it at both levels: Congress and your state. A lot of AI law is happening in states right now.
  • Do it again in a month. Repeat contact is what gets noticed.

Show up in person

Face to face is the strongest thing you can do. Most of it is easier than it sounds.

  • Town halls. Lawmakers hold these in their districts, often during breaks like August, holidays and the weeks before an election. To find them, sign up for their email newsletter and follow them on social media, since that’s where they announce. Ask one of the questions from the email above.
  • Meet at their local office. Every member of Congress, and most state legislators, has an office near you. Use the “request a meeting” form on their website. You’ll usually meet a staffer, and that counts. Bring a friend or two and your one most important ask.
  • Testify or comment at a state hearing. Your state legislature’s website lists committee hearings on upcoming bills. Many let the public speak or send written comments, and some allow remote testimony.
  • City council and county meetings. Agendas go up on your city or county website a few days ahead. Most meetings have public comment, usually two or three minutes. This is where data center decisions get made.
  • Candidate events before November 3. Right now candidates are doing forums and meet and greets. It’s the easiest time to ask them on the record where they stand.
Questions to ask candidates →

Who your reps hear from

AI and tech companies spent about $41 million lobbying Congress in the first half of 2026. That’s over $226,000 a day. [21]

They spend on elections too. The AI industry’s biggest super PAC network spent $8.2 million to defeat Alex Bores, the New York lawmaker who wrote his state’s AI safety law. [22] [23]

But money keeps losing on this. When the industry pushed to block state AI laws, the Senate voted 99 to 1 against it. [24] That super PAC network reported zero new donations last quarter. And the person who beat Bores supports tough federal rules for AI companies too. [23]

Lobbyists have money. You have a vote in their district. That’s why your message counts.

What we’re asking of AI companies

Open +

To OpenAI, Anthropic, Google DeepMind and xAI:

  • Give independent testers access before high risk releases.
  • Tell the public about serious incidents.
  • Protect employees who raise safety concerns.
  • Make sure dangerous systems can be restricted or shut down, and cloud providers can carry out lawful emergency restrictions.

These are our demands, not promises the companies have made. Researchers build the evidence, officials set rules that can be enforced, companies follow them, and the public keeps everyone accountable. Nonprofits can test and push, but they can’t replace a regulator with real legal power.

Who’s fighting for workers

Open +

Jobs are where AI rules are weakest. These are the people actually pushing:

  • Unions. The AFL-CIO, the biggest group of U.S. unions, has AI principles to protect workers. [25] The NewsGuild has AI rules in about 85 to 90 contracts, SAG-AFTRA stopped studios from replacing actors with AI copies without consent, and the union at video game company ZeniMax won a say before new AI tools are brought in. [26]
  • States.
    • California: since September 2026, employers must say when AI causes a mass layoff, and a real person must review AI firing or discipline decisions. [27]
    • Connecticut: since October 2026, employers must say whether AI played a part in a mass layoff. [28]
    • Illinois: since January 2026, employers must tell workers when AI is used in hiring, firing or discipline, and can’t use it to discriminate. [28]
    • Texas: since January 2026, employers can’t use AI to intentionally discriminate in job decisions. [28]
    • New York: layoff notices now ask whether automation played a part. In the first year, not one company said yes, which shows why clearer rules matter. [29]
    • New Jersey: a pending bill would require 90 days’ notice before AI layoffs of 25 or more people, plus extra unemployment benefits. [30]
  • Congress, from both parties. Sen. Mark Warner (D, Virginia) and Sen. Josh Hawley (R, Missouri) introduced a bill to make companies report jobs lost to AI. It hasn’t passed yet, so ask your Senators to back it. [31]

Beyond your reps

Open +

Laws lock in change, but pressure from the ground up is what gets them passed. Three other ways in:

  • At work: bargain over how AI is used in your job. [26] [25]
  • With your wallet: choose which companies get your money and your data. [32]
  • In court: lawsuits can hold companies responsible for harm. [33]

Groups doing the work

Open +

Who tests AI

METR is a nonprofit that tests the most advanced AI models for dangerous abilities, like carrying out cyberattacks or being hard to shut down, and shares what it finds with the public. It’s one of the few outside checks we have, so it’s a good place to see what the newest systems can actually do. [34]

See 5 groups
Done? Now tell one person you did it →