AI policy in the US government

Why work in AI policy within the US government?

In a recent poll, 65% of Americans said that the government has done too little to regulate AI. But wanting more regulation and getting useful regulation are different things. In the 2010s, voters broadly supported federal action on the opioid epidemic, and Congress responded with major legislation in 2016 and 2018. These laws had some good provisions, but didn’t stop the crisis. Opioid deaths kept rising through 2022, in part because the threat evolved: the laws primarily addressed opioid prescriptions, and the increase was driven by illegally produced fentanyl.
The risks of AI will continue to evolve: the regulation that seems most obvious today may not sufficiently address the threats we’ll face in a year or a decade, and the stakes are too high to get this wrong. We need talented, thoughtful people working to make AI regulation actually work.

The US federal government is likely to be the most consequential regulator of AI in the world. It has jurisdiction over the most prominent AI companies — Anthropic, OpenAI, Google DeepMind — as well as key parts of the chip supply chain, and its decisions about whether and how to regulate frontier AI development will shape the technology’s global trajectory. Because decision makers often defer to experts (like agency employees or congressional staff), even people in relatively junior roles can make a big difference.

A lot of AI policy work happens upstream of government, in research organisations and advocacy groups. But ideas only matter if someone in government decides to make them happen, whether it’s a member of Congress working them into a bill or a regulator turning them into enforceable rules. Implementation — getting the details right, navigating interagency politics, drafting language that holds up — is often a bigger bottleneck than coming up with good ideas in the first place.

Would you be a good fit?

What skills are needed to succeed?

Most government AI policy roles call for some combination of:

• Subject matter expertise. Knowing the basics of how AI systems actually work, what current policy proposals are, and where the technical and political fault lines lie. You don’t need to be a frontier ML researcher, but you need enough fluency to engage seriously with the technology and the debates around it.
• Institutional knowledge. Understanding which agencies do what, who the key staffers are, how bills are passed and rules are established, and what’s politically possible at any given moment. A lot of this is learned on the job, but starting with some grounding helps you earn trust and avoid mistakes.
• Relationship-building skills. The government runs on networks. You’ll need to build trust with colleagues, brief people clearly, follow up reliably, and navigate environments where decisions often get made through informal conversations rather than formal processes.

That last point is worth emphasising. Politicians and policymakers don’t have time to pick up expert knowledge on all the issues they deal with; they rely heavily on people they trust to help them develop views. Becoming one of those people is partly about knowing your stuff and partly about being someone others like and trust.

It’s also worth being honest with yourself about whether you’d thrive in the political environment. Washington has a particular culture: it’s heavy on networking, status, and influence. Merit — having the right ideas or delivering results — isn’t the only driver of success. People who find that culture draining or off-putting should think carefully about whether this is the right path, or whether a different role in AI policy (such as research or technical governance) might be a better fit.

What experience is useful

People take many different paths to reach the US government, but most have one or more of the following kinds of experience:

• Relevant academic degrees. Policy master’s degrees are particularly useful in DC, as are law degrees and PhDs in relevant fields. They’re often required or strongly preferred for senior roles, and they can let you enter the bureaucracy at a higher level.
• Jobs that demonstrate subject matter knowledge. Work in industry, technical research, or other AI-relevant fields can establish you as someone who actually understands the technology. This is often more valuable than a generic policy background.
• Prior government or political experience. Time in any part of government — a congressional office, a federal agency, a state legislature, a campaign — gives you institutional knowledge and a network that’s hard to build any other way.

What else do you need?

Government policy roles are typically in-person office jobs, and tend to be more formal and professional than most roles in tech or academia. They also tend to be highly collaborative: it’s rare for policy to be created entirely by one person, and many of the people you’ll work with will have different worldviews and styles of thinking. If that way of working appeals to you, this career may be a good fit.

For most federal roles, you’ll need US citizenship, and you’ll need to be willing to live in DC or spend a lot of time there. If you’re not a US citizen, your options are more limited — many think tanks are open to non-citizens, and some congressional staffing roles are accessible to green card holders, though it makes those jobs even harder to get. For more on what’s available, see Emerging Tech Policy Careers’ guide for foreign nationals.

A lot of these jobs are also partisan. If you have prior associations with one party — including donations — you’re effectively locked into that party for many roles, and may be locked out of political appointee roles while the other party holds the Oval Office.

Downsides to working in US government AI policy roles

Risk of doing harm

The US government is powerful, and even relatively small government policies can have a big impact. This could position you to do a lot of good — or a lot of harm.

In government, you’ll primarily be working with people who aren’t prioritising AI safety, so it may be difficult to keep your work focused on that goal. Also, the government tends to move slowly and resist change, which makes it harder to fix mistakes (reversing a policy that turns out to be harmful is much harder than correcting a misguided think tank report or changing a corporate practice).

You might accidentally cause harm by:

• Politicising AI safety by associating it with a particular ideological camp, making it easier for opponents to dismiss
• Sending the message, implicitly or explicitly, that the risks are being managed when they aren’t, or that they’re lower than they in fact are
• Suppressing technology that would actually be extremely beneficial for society

While you can never be certain that your work will have these effects, you should keep them in mind as you embark upon this career — and remain open to changing course if you find evidence that your actions may be damaging.

We also recommend the following guidance:

1. Ideally, eliminate courses of action that might have a big negative impact.
2. Develop expertise, get trained, build a network, and benefit from your field’s accumulated wisdom.
3. Match your capabilities to your project and influence.

Risk of burnout

Government work can be unusually prone to burnout. It often involves long hours, low job security, and high turnover; you may find yourself frequently rebuilding relationships and adjusting to new colleagues and bosses.

Top organisations

In the US government, AI policy work spans the executive branch, Congress, state governments, and a network of contractors and outside organisations that work closely with all of them. We’ve listed some of our top recommendations below, but the institutional landscape shifts constantly. Departments restructure, administrations change, and elected officials retire; the people who matter most for AI policy this year might not be the same people next year. This is part of what makes government work both interesting and demanding — but it also means specific recommendations about agencies and roles age quickly, and you’ll want to stay plugged into current conversations rather than relying on any single source (including this one).

Executive branch

Executive Office of the President

The White House and the Executive Office of the President house a number of offices that touch directly on AI policy. The most relevant include:

• The Office of Science and Technology Policy (OSTP), which advises the President on science and technology issues, including AI.
• The Office of Management and Budget (OMB), which oversees federal agencies’ budgets and regulations and has played a significant role in shaping how the executive branch uses and procures AI.
• The National Security Council (NSC), which coordinates national security policy across agencies and has increasingly focused on AI as a national security issue.

These roles are among the hardest to obtain and often go to people with long-standing relationships in government or significant tenure in relevant areas. They can be entered through senior roles in other parts of the executive branch, fellowships (such as the Presidential Innovation Fellowship or Presidential Management Fellowship, listed below, or political appointments.

Departments and agencies

Agency roles split roughly into two categories:

Political appointees are chosen by the White House and serve at the pleasure of the President. These roles tend to be more impactful than civil service roles — they shape strategy and direction rather than just implementing existing rules — but they almost always require affiliation with whichever party currently holds the White House, and they turn over with each administration.

Civil servants are career staff who stay across administrations and focus on implementing policies. While senior civil service roles are valuable, we generally don’t recommend taking a low-level civil service job as a first step towards AI policy impact — it usually takes a long time to get a senior role. If you already have a lot of relevant career capital outside of government, or are already a relatively senior civil servant, civil service roles could be a good path to impact for you.

Departments and agencies especially likely to be relevant to AI policy include:

• Commerce, which houses both the National Institute of Standards and Technology (NIST) and the Bureau of Industry and Security (which administers the chip export controls central to current US-China AI policy)
• Defense, though there’s also significant potential for harm here, and we haven’t recommended many DoD roles
• Energy, which oversees the national labs that house major US supercomputing infrastructure
• State, particularly for AI-related diplomacy and international coordination
• The intelligence community, which includes work on cybersecurity, global AI power dynamics, and monitoring dangerous uses of technology

Congress

In Congress, you can work directly for individual lawmakers as part of their personal staff, or for legislative committees as committee staff. Personal staff handle the full range of issues a member’s office deals with — constituent services, scheduling, communications, and a portfolio of policy areas. Committee staff specialise in the policy areas under their committee’s jurisdiction, and they tend to be more deeply involved in actually writing legislation. Committee staff roles are generally more prestigious and more influential on AI policy specifically, but they’re also more competitive. A common path is to start as a personal staffer for a member who’s active on AI issues and transition to a committee role once you’ve built up experience and a network.

Roles in Congress are pretty much always partisan. Working for the majority party is generally more impactful, but if your party is in the minority, the role can still be great career capital — and you might find yourself in a better position after the next election.

Some committees are likely to be more involved in AI policy than others. The ones we’d flag as worth targeting include:

• Senate Committee on Commerce, Science, and Transportation — handles AI research, industry standards, and US competitiveness with China; this committee has been particularly active on AI legislation in recent years
• Senate Judiciary Committee — takes up AI harms, safety, deepfakes, and algorithmic bias when these come before Congress
• House Energy and Commerce Committee — covers commercial AI regulation and consumer protection legislation
• House Committee on Science, Space, and Technology — has jurisdiction over emerging tech policy, plus the NIST, NSF, OSTP, and DOE national labs, which makes it especially relevant for compute regulation and research-related AI policy
• House Judiciary Subcommittee on Courts, Intellectual Property, Artificial Intelligence, and the Internet — despite its name, this subcommittee’s AI work is mostly IP-focused (copyright disputes over training data, patent issues for AI-generated work), with occasional broader AI hearings
• Armed Services Committees (House and Senate) — oversee military uses of AI, including autonomous weapons, defence procurement, and cyber and intelligence systems used in the military
• Intelligence Committees (House and Senate) — cover AI competition with China, cyber intelligence, and AI applications in national security
• Appropriations Committees (House and Senate) — control funding for AI research, defence and civilian AI programmes, and agency budgets, making them more impactful for AI policy than people often realise

State governments

Federal AI policy gets the most attention, but state governments are increasingly important. The advantages of state-level work include:

• State policy can be important on its own and can shape national conversation.
• States move much more quickly than the federal government.
• State roles are less competitive and the field is less saturated.

There’s a risk that federal action — either specific preemption or a broader regulatory framework — could override state-level work. But state policy, especially in the largest states, can be important enough that we think it’s worth considering.

The most impactful state roles tend to be in CA, NY, TX, FL, and WA in legislatures, governors’ offices, or agencies that implement relevant policies. Emerging Tech Policy Careers has more on state-level pathways.

Government contractors

A meaningful amount of government work — especially research and analysis — is contracted out to outside organisations. RAND is the most well-known example, but there’s a broader ecosystem of contractors and federally funded research and development centres (FFRDCs) doing technical and policy work for federal agencies. Although these aren’t part of the government, many people who spend their careers in government spend time at these organisations, whether early on (to build career capital), when their party is out of power, or when they see a particularly beneficial opportunity. It’s common to move between contractors and government throughout a career.

Junior roles at think tanks can offer particularly good career capital for people who eventually want to work in Congress or the executive branch. They give you a chance to develop expertise, publish, build a network in DC, and demonstrate that you can do policy-relevant work.
For more on the most important places to work within and adjacent to government, see our article on the US AI policy landscape.

Next steps

If you’re ready to apply for jobs

Our job board features AI policy opportunities in the US government:

View all opportunities

If you need to build career capital

Become known as someone who knows about policy. This means writing online about AI policy questions, going to relevant conferences and events, meeting people, and putting yourself in conversations where people in the field will encounter you.

Get credentials that establish your AI expertise. Jobs at AI companies, technical degrees, work on AI-relevant research, courses, and fellowships all count. You don’t need to be a leading researcher — you just need to credibly demonstrate familiarity.

Get good at networking. A lot of policy work depends on talking to people. 20–30 informational coffee chats isn’t unusual when you’re trying to break into a new corner of DC or figure out what role might fit you. Emerging Tech Policy Careers has a useful guide to networking in this space.

Consider working on campaigns. Working on a campaign can be a path to a congressional staffing job if your candidate wins, especially with the midterms coming up. See our article on US electoral politics for more.

Apply to fellowships and internships. Fellowships in particular are among the best entryways into policy work — they offer first-hand experience, funding, training, mentoring, and a network. Some are open to college graduates; many require an advanced degree. Here are a few we recommend:

• Presidential Management Fellowship
• Presidential Innovation Fellowship
• Horizon Fellowship
• TechCongress Fellowship
• STPI Science Policy Fellowship
• AAAS Science & Technology Policy Fellowships

Increase your visibility. Getting a job is just the beginning. Once you’ve been hired, you can become much more influential by giving input on draft reports, comment periods, and internal working groups, scoring invites to meetings where powerful people are present, building a reputation as someone worth consulting — so that over time you’re in the room for bigger decisions and better-positioned for senior roles.

Learn more

Top recommendations

• BlueDot Impact’s selective Frontier AI Governance course can be taken full time (six days in total) or part time over a longer period so that it fits alongside your job or studies. It covers the frontier AI landscape, key governance debates, and how to find your own path into the field, and is aimed at people seriously considering AI governance work. Alumni have gone on to work at places like RAND, CSET, IAPS, GovAI, CAISI, and frontier lab policy teams.

• Emerging Tech Policy is a free career resource website run by the Horizon Institute for Public Service. It covers pathways into policy, profiles of key institutions (Congress, think tanks, federal agencies), and guides to specific policy areas including AI, biosecurity, and cybersecurity. It’s aimed at people exploring or actively pursuing careers in tech policy, and is particularly useful for understanding how different institutions work, where you might fit, and what you might want to read next.

Further recommendations

• Career review: US electoral politics
• Career review: AI safety advocacy
• Podcast: Dean Ball on how AI is a huge deal — but we shouldn’t regulate it yet
• Podcast: Paul Scharre on how AI could transform the nature of war
• Podcast: Eileen Yam on how we’re completely out of touch with what the public thinks about AI
• Podcast: Nathan Calvin on California’s AI bill SB 1047 and its potential to shape US AI policy
• Podcast: Carl Shulman on the economy and national security and government and society after AGI
• Podcast: Lennart Heim on the compute governance era and what has to come after
• Podcast: Vitalik Buterin on defensive acceleration and how to regulate AI when you fear government
• Podcast: Tantum Collins on what he’s learned as an AI policy insider

Acknowledgements

We thank Randan Steinhauser for feedback on a draft of this article.

Read next:  Learn about other high-impact careers

Want to consider more paths? See our list of the highest-impact career paths according to our research.

Continue →