A way to help win AI safety: Staff the new Congress
TL;DR: The midterms elected a large number of new lawmakers. They are hiring staff over the next couple months. Their tech policy staff will have significant influence on AI policy that happens in the next year (which will be eventful). High-context people in AI safety should try to get those jobs. The AI safety movement should collaborate to help get them chosen.
If you are persuaded and decide to apply for any of these roles, tell us by filling out this form: AI safety staff applicants for the new Congress. To help you do so, read this guide: A Guide for AI Safety Applicants.
Tell us you applied Read the guide
Who this appeal is aimed at
You're someone who is worried about the risks of advanced AI: loss of human control over super intelligent AI systems, AI-enabled biological or cyber weapons, and AI used to concentrate power or build authoritarian surveillance states, etc. You also think the companies building these systems are not doing enough, and that binding rules are needed - rules that can only come from government action.
You know a lot about AI. You're "high-context." You have read the core arguments and papers, maybe you've attended a field-building intensive like BlueDot, and you have opinions about the potential policy solutions.
You are worried about the next two years. You want to do something about it, but are one of the many capable people who the AI Safety field has not been able to absorb. Or maybe you are working in a research or technical role, but aren't sure if you are making enough of a positive impact upon the trajectory we are on.
I urge people like you to consider getting a job working for Congress, or for one of the other elected officials newly entering office following the 2026 Midterms.
It's clear to a lot of people like me who work on advocacy and policy related to AI safety that this is an important and impactful thing for more people to do. After the election, Charlie Bullock of the Institute for Law & AI tweeted to urge people in AI policy to try to work as congressional staffers, both junior ones as well as senior people. In June, Transformer published a piece about how "building independent expertise inside Capitol Hill could reduce reliance on industry briefings," which would result in faster and better outcomes.
Staff shape what members think, and a lot of members have unformed views right now
The staff who advise members of Congress on AI over the next two years will shape what the government does about it during what is likely to be a crucial window.
A member of Congress votes on countless issues, while having direct expertise in very little of it themselves. Each relies on a small staff: often 12-18 people in a House office, divided between DC and their district. In most offices one single staff member known as a legislative assistant covers technology (including AI), within a portfolio including other policy areas.
That staffer recommends how the member votes and which bills to sponsor. They come up with and evaluate ideas for bills and amendments, write briefing memos and talking points, prepare the member for hearings, and they take meetings with outside experts, advocates, constituents, and lobbyists on their boss's behalf.
Committees do the same work in a deeper and more comprehensive way, working primarily for the chair (majority party, who has a much larger staff) and ranking member (minority party) of the committee. Their staff prepare the hearings and propose who the witnesses should be, lead investigations and briefings, draft and review major bills, and each member's own staff coordinate with the committee staff to develop legislation reflecting the member's priorities.
On an emerging issue or subject the member has not studied closely, or which they aren't hearing about from their constituents, the staffer's view is often very influential. Hill staff describe experiencing “Hey, my boss wants to do something on AI. That’s all the direction I’ve gotten” (Transformer).
Today, by default the person filling these roles is usually a capable generalist, but with no background in AI or understanding of AI risks. Where Congressional staff lack technical knowledge, company-sponsored briefings and industry lobbying fill the vacuum. It is also a challenge to hire people with expertise in AI, because that expertise is in high demand and Congress cannot match what companies pay.
Yet, who fills these jobs over the next two years is likely to shape what Congress does about AI during the period of time when many experts expect a rising number of "warning shots," the labs achieving full automation of AI R&D, and a potential intelligence explosion leading to decisive path dependency for humanity's future.
A wave election opened these jobs all at once
More than 100 members of the House of Representatives, 18 members of the Senate, and 21 Governors are newly elected, and will be filling their staff mostly from scratch over the next two to three months. The newly-elected Congress is sworn into office in January, 2027.
Relative to the currently-seated Congress sworn in after the 2024 election, roughly a quarter of the House and a fifth of the Senate will be new faces.
A preliminary estimate by POPVOX predicts a hiring wave of about 3,650 staff positions across the House and Senate between November 2026 and March 2027: about 2,530 in the House and 1,120 in the Senate.
Several things make the next three months an unusual window of opportunity:
- New offices are starting empty. A House office typically employs up to 18 permanent staff, typically split between policy staff in D.C. and casework/community engagement in their district. A new senator hires 30 to 40 people split between their state and D.C. This means a lot of hiring, all at once, right now.
- Committees changed hands. Democrats won both chambers, flipping control of the committees from Republicans to Democrats, so new chairs will rebuild the majority committee staffs by expanding their existing teams. The committees that oversee AI will be looking in particular to increase their in-house expertise as part of this reshuffling.
- More seats on AI committees. Because of the size of the Democrats' new majority, the ratio of Democrats to Republicans on each committee will be higher than normal, meaning that even very powerful and desirable committees with jurisdiction over AI (like Energy & Commerce) will have more seats than expected for additional Democrats to join, increasing the number of members with major influence over AI policy.
- AI is an unclaimed issue portfolio in many offices. As of October, no House member appeared to employ a staffer dedicated solely to AI (Transformer), with the exception of Rep. Nathaniel Moran who recently created an AI Policy Fellow role. In most offices it is one of several issues in someone's portfolio. Even just the typical "technology" portfolio encompasses a wide range of issues beyond AI. There is an opportunity for you to stake a claim on AI on the basis of the context you'd bring with you.
Offices hire a skeleton staff first, often the most senior roles like Chief of Staff, and fill legislative roles after committee assignments have been finalized, a process which may be prolonged into February or later.
Someone will fill these jobs, and better if it's us
These jobs will be filled by spring whether or not anyone from AI safety applies. They could be filled by people who think AI risks are fake and "hype," or people who are inclined to trust the industry lobbyists.
I believe that someone who cares about AI safety taking one of these jobs can do a great deal of good, but if MANY people who care about AI safety take these jobs then the impact will be even greater, creating a beneficial environment of consensus and urgency in Congress, and having tractable and positive policies worked out and ready the moment a political window opens.
It's worth acknowledging reasons people reading this might hesitate:
- The pay is low and the hours are long. A legislative assistant earns roughly $60,000. What you get in exchange is a hand in what the government does about AI during a crucial window.
- You will work for one member of one party. You will be expected to represent and defend that member's record on every issue, not only AI. Apply to work for members you are able to loyally represent.
- The portfolio will not be only AI. Technology is usually bundled with other issues. Covering them is the price of holding the AI portfolio.
- Your words and public record may be exposed to scrutiny. If you like to post on forums or Twitter under your real name, at minimum you will need to be mindful of your association with your boss going forward.
- The roles are competitive. This is why the AI safety movement should collaborate to help get its people chosen, and work together to draw upon political allies and connections to make referrals.
- You'll be working for just 1 member of 435 (or 1 of 100 in the Senate), how impactful could that be? It's true that just one vote in Congress is just one vote, but getting the marginal additional member of Congress to care about AI risks has a bigger impact than you might think, and building enough support among individual members can cascade into bringing a policy from marginal to consensus quite fast.
What to do next if you are persuaded
Apply to roles asap, and reach out so that allies can help you.
- Read this guide. Staffing the new Congress: a guide for AI safety applicants is a living document which has the steps to take, the eligibility rules, the roles and what they pay, and every resume bank, member office application link, and opportunity we have found.
- Apply. Put your resume into the official House and Senate collections and the pipelines that match your politics.
- Tell us you applied. Fill in this form so that people who know those offices can recommend you, or we can provide you advice or match you to other opportunities that might also be a good fit.
If you know people in Congress or who work in these offices and can help, reach out to me at amccoy@safeai.org.
Who's behind this
I'm Alexander McCoy (@AlexanderMcCoy4 on Twitter), and work in AI safety political advocacy, currently with Citizens for Safe AI for this sprint effort, and previously at AI safety advocacy groups Irreplaceable, and Humans First. I've also worked on AI tech policy as a legislative staffer myself, for the chair of the NY state senate's tech committee. In the 2020-2021 transition, I co-chaired a coalition of human rights NGOs recommending 100 foreign policy-related personnel appointments to the Biden-Harris admin, and we were able to get about a quarter of them selected, including for some senior roles.