About & Methodology

What this is

AI on the Ballot documents the publicly available AI governance positions of U.S. congressional candidates. We track what candidates have said, how they have voted, and what legislation they have sponsored or cosponsored on key AI policy issues.

This is a nonpartisan transparency resource. We do not evaluate, endorse, score, or make recommendations about candidates or electoral outcomes.

Methodology

Every candidate is researched across four source categories:

• Campaign websites. Official policy pages, press releases, and issue statements.
• Social media. Public posts on major platforms where candidates discuss AI policy.
• Web search. News interviews, op-eds, debate transcripts, and public remarks.
• Congressional record. Sponsored and cosponsored bills, committee hearing statements, and floor votes for incumbents.

How stances are coded

Each candidate’s position on each tracked issue is coded into one of five categories:

• Support. Candidate has clearly expressed support for the policy area or approach.
• Oppose. Candidate has clearly expressed opposition to the policy area or approach.
• Mixed. Candidate has expressed both supportive and opposing views, or supports some aspects while opposing others.
• Unclear. Candidate has addressed the topic but their position cannot be confidently categorized.
• No mention. No public record of the candidate addressing this topic was found in our research.

How confidence is assigned

Each coded stance carries a confidence level reflecting the strength of the underlying evidence:

• High. Based on direct, unambiguous statements or legislative actions, e.g., a sponsored bill, explicit policy page, or floor vote.
• Medium. Based on clear but indirect evidence, e.g., social media posts, interview remarks, or cosponsorship of a related bill.
• Low. Based on inferred or tangential evidence, e.g., a brief remark in a broader context, or a position extrapolated from related statements.

Tracked issue categories

We track candidate positions across the following AI governance issue areas:

• Export Control & Compute Governance. Positions on restricting chip exports, controlling access to advanced compute, and semiconductor policy.
• Military & National Security Uses of AI. Positions on AI use in defense, autonomous weapons, intelligence applications, and national security frameworks.
• AI Regulation Philosophy. Positions on how (or whether) to regulate AI: licensing, liability, open-source, and federal agency roles.
• AI Companion Chatbots. Positions on AI companions, romantic or emotional companion chatbots, and their psychological effects on users.
• Children's Online Safety. Positions on AI-driven content moderation for minors, age verification, and algorithmic protections for children.
• Data Centers. Positions on data center permitting, energy consumption, federal support, and environmental impact.
• Jobs and Workforce Disruption. Positions on labor market effects of AI: automation, displacement, retraining, and workforce policy.
• Deepfakes and AI Fraud. Positions on AI-generated impersonation, synthetic media, election deepfakes, and related fraud enforcement.
• AI Preemption. Positions on whether federal AI law should preempt state and local AI regulation.
• Intellectual Property and AI. Positions on copyright, training-data licensing, and author/artist protections against generative AI use.

Candidate inclusion criteria

Candidates must meet FEC filing and fundraising thresholds to be included:

• Senate. FEC Form 1 (Statement of Candidacy) filed, plus $100,000 or more raised.
• House. FEC Form 1 filed, plus $15,000 raised.

Thresholds are re-evaluated at each FEC quarterly filing deadline. Candidates who fall below the ongoing threshold may be removed from active tracking.

Candidate outreach

As part of our methodology, we are reaching out to each candidate appearing in this tracker to share how their AI governance positions are being characterized on this site, with citations for the public sources used in each coding decision. This outreach is rolling: candidates are contacted as their state’s data comes online, so not every candidate listed will have been contacted at the time of publication.

The outreach is intended to flag any mischaracterization before or shortly after publication, and to invite the candidate or their campaign to clarify their stance, either by pointing us to additional public sources or by submitting a formal clarification through our feedback form.

Where a candidate has not yet been reached or chooses not to respond, we publish based on the public record. A candidate’s silence is not interpreted as agreement or disagreement with how a position is coded.

Coverage timeline

Candidate data is released on a rolling state-by-state schedule, aligned with each state’s primary calendar. At the latest, data for a given state is released three days ahead of that state’s primary. Key Senate races are noted in parentheses.

Updates and corrections

The candidate dataset is updated on the FEC quarterly schedule: April 15, July 15, and October 15. Position coding is reviewed monthly between quarterly updates.

If you believe any information on this site is inaccurate, incomplete, or outdated, submit a correction. All submissions are reviewed before any changes are made.

Editorial Policy

AI on the Ballot is a nonpartisan research project tracking congressional candidates’ public positions on AI governance issues. This project originated within MATS Research and is now a project of Evitable. The AI on the Ballot research team operates with full autonomy over candidate selection, data collection, coding decisions, and research methodology. These decisions are made independently of Evitable’s other activities.