Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

Compare candidates on AI by checking the same specific policy questions for each person, then tying every conclusion to a dated public statement or recorded action. A missing statement is not evidence of support or opposition. The records and examples below focus on U.S. federal candidates; for state and local races, use the election, legislative, and disclosure offices in the relevant jurisdiction.

Start with specific AI policy questions

Do not reduce a candidate to “pro-AI” or “anti-AI.” Those labels conceal differences between issues and can make an unsupported impression look like a researched position. Set an issue list before reviewing candidates, and use the identical list and order for everyone.

Useful dimensions include:

  • AI regulation and oversight
  • Compute, chip access, and export controls
  • Military and national-security uses
  • Children’s online safety
  • Data centers, infrastructure, and energy
  • Jobs and workforce disruption
  • Deepfakes, impersonation, and AI-enabled fraud
  • Federal preemption of state rules
  • Intellectual property and the use of copyrighted material

These are separate questions, not a single scale. A candidate could support one kind of oversight while opposing another, or discuss workforce effects without addressing data centers. Issue categories used by AI on the Ballot’s methodology offer one example of this issue-by-issue approach.

Find the right public records

For federal candidates

The Federal Election Commission’s candidate research guide points to candidate lists, campaign-finance summaries, filings by candidates and authorized committees, related committee filings, and itemized contributions. It also directs readers to state offices that maintain candidate reports and disclosures.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

For a state or local race, first identify the jurisdiction. Consult its election office for candidate and election information, its legislature for bills and votes when applicable, and the appropriate disclosure agency for campaign records. The FEC guide links to state disclosure resources, but state records and requirements vary.

Separate policy records from finance records

Campaign-finance records can establish facts about filings, receipts, contributions, disbursements, and committee activity. They do not, on their own, establish a candidate’s AI policy. Look for policy evidence in candidate-controlled issue pages, official statements, questionnaires, interview or debate transcripts, and—in the case of incumbents—bills, committee statements, and recorded votes.

Build a source ledger for every claim

Keep a simple record for each issue and candidate. This makes it possible for another reader to check how you reached a conclusion and prevents a paraphrase from drifting beyond what the source says.

  • Issue: the specific AI policy question.
  • Source: a direct link to the page, document, transcript, or recording.
  • Evidence: the exact relevant words or the recorded action, with enough context to understand it.
  • Date: when the statement or action occurred or was published.
  • Source type: for example, campaign policy page, interview, bill sponsorship, or vote.
  • Limit: what the evidence does not establish, such as whether a narrow bill reflects a broader policy position.

Prefer the candidate’s own words and official legislative records where available. If a news report summarizes an interview, link to the transcript or recording when you can. VoteVera’s published methodology describes a verification approach that prioritizes recorded votes and candidate-controlled sources over interviews and press quotations, and calls for a linked, dated quotation when coding a stance.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

Label the evidence without overstating it

A statement, a bill, a cosponsorship, and a vote are different kinds of evidence. Describe the source accurately, then state only what it supports. A vote may show how a candidate acted on a particular measure; it does not automatically reveal a complete philosophy of AI policy. Sponsoring or cosponsoring a bill likewise supports a claim about that measure, not necessarily every related policy question.

Use clear, non-numeric labels such as:

  • Direct and clear: a dated statement explicitly addresses the issue.
  • Indirect: an action or broader statement is relevant but does not directly answer the question.
  • Mixed: the candidate supports one element and opposes another, or has made conflicting statements.
  • Unclear: available evidence does not support a confident interpretation.
  • No public position found: no relevant public evidence was located in the sources reviewed.

“No public position found” is a finding about the records you checked, not proof that the candidate has no view. Do not infer a position from party affiliation, donors, a related issue, or a single broad remark. Keep the underlying quotation or action visible rather than hiding it behind a score. AI on the Ballot and VoteVera both describe uncertainty and evidence quality as part of their methodologies (AI on the Ballot; VoteVera).

Account for dates, qualifications, and changes

Attach a date to every finding. If a candidate later makes a clear statement that updates an earlier one, include the newer evidence and explain the change where relevant. Older votes and statements remain part of the record, but may not represent the candidate’s current position.

When a candidate’s view is qualified, report both sides rather than compressing them into a yes-or-no label. For example, distinguish support for a particular disclosure requirement from a broader claim about regulating AI-generated content. Public trackers warn that their records may be incomplete and that positions can change; their summaries are useful leads, not substitutes for the linked records.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Support on Ko-Fi

Present the comparison so readers can check it

Use the same issue order and evidence standards for all candidates. A table can make gaps and distinctions visible without suggesting that every row has a definite answer.

Issue Candidate A Candidate B
AI regulation Position summary; linked statement or action; date; evidence label Position summary; linked statement or action; date; evidence label
Data centers and energy Position summary; linked statement or action; date; evidence label Position summary; linked statement or action; date; evidence label
Deepfakes and fraud Position summary; linked statement or action; date; evidence label Position summary; linked statement or action; date; evidence label
Workforce and jobs Position summary; linked statement or action; date; evidence label Position summary; linked statement or action; date; evidence label

These are example rows, not findings about particular candidates. In a real comparison, replace the example labels with sourced evidence. Mark an issue “not researched” if you have not checked it, and use “unclear” or “no public position found” only after explaining the scope of your search. If a tracker covers only selected candidates or issues, state its selection criteria and coverage so readers know what its comparisons include. The AI Congress tracker is an example of an issue-based comparison that makes coverage limits visible.

Keep legal context in its proper time frame

Legal claims about AI in campaign advertising can become outdated. The Congressional Research Service report Artificial Intelligence (AI) and Campaign Finance Policy: Recent Developments, updated August 27, 2024, states: “No federal statute or regulation specifically addresses artificial intelligence (AI) in political campaigns.” That is the report’s dated assessment and should not be presented as a statement of law in October 2026 without checking current statutes, regulations, and agency materials.

The same 2024 report discussed how federal campaign-finance disclaimer rules appeared to apply to AI-generated ads without requiring an alert that an ad was AI-generated, alongside proposals and agency activity in 2023–2024. Treat those details as historical context, not current legal guidance.

Free tools Windows power users keep installed

One-click scans. No signup required.

Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.

Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.