Overview
1. Which AI vendors are campaigns paying?
Every disbursement's payee name, stated purpose, category description, and memo text is matched against a taxonomy of both general-purpose AI vendors (OpenAI, Anthropic, Google, Microsoft, xAI, Perplexity, and AI-native creative tools) and vendors built specifically for campaigns. The taxonomy started from vendors named in press coverage (Quiller, BattlegroundAI, Chorus AI, DonorAtlas, RivalMind, Aaru), then was expanded by scanning the disclosures themselves for AI-indicative language and researching the payee names that kept turning up. See methodology for the full list, its confidence tiers, and the vendors we considered and rejected. Several of the biggest names by dollar volume — Amplify.ai, Prompt.io, CallTime.AI — are companies founded years before ChatGPT existed and market themselves as “AI” the way that word was used before the current generative-AI wave; they're tagged as legacy vendors and hidden from charts and tables below by default (toggle at the top of the page to include them).
AI-related disbursements by vendor, all committees, 2020–2026 cycles
High-confidence name matches only. Bars colored by whether the vendor is general-purpose or built specifically for political campaigns.
Vendor co-occurrence
Normalized as Jaccard similarity: of the committees that pay either vendor, the share that pay both. Top 15 generative-era vendors by high-confidence spend, by committee (not candidate). See methodology.
2. What are candidates reporting they use AI for?
Disbursement purpose/category/memo text for House and Senate candidate committees, clustered into use-case categories by keyword. A single disbursement can land in more than one category (e.g. a memo mentioning both an email platform and ad creative), so these are not mutually exclusive shares of one total.
Spending by stated use case
Candidate committees only, high-confidence vendor matches.
Use case, by vendor type
Same disbursements, split by general-purpose vs. campaign-specific vendor.
3. How does AI use differ across candidates?
2026-cycle House and Senate candidate committees with a high-confidence AI vendor match, split by party, incumbency status, candidate age, and chamber. Amounts are totals; candidate counts are shown alongside since a handful of high-spending campaigns can otherwise dominate a total.
By party
2026 cycle.
By incumbency status
2026 cycle.
By chamber
2026 cycle.
By candidate age
2026 cycle. Age is known mainly for incumbents/former members (see methodology) — "Unknown" is shown rather than guessed.
4. What's changed across cycles?
The same candidate-committee, high-confidence disbursements, tracked across the 2020, 2022, 2024, and 2026 two-year cycles.(Mainstream generative-AI tools (ChatGPT and peers) were not publicly available for most of the 2020 and 2022 cycles.)
Total AI-related spending by cycle
General-purpose vs. campaign-specific vendors.
By party, over time
By incumbency, over time
By chamber, over time
Weekly disclosure timeline
Every matched disbursement, all confidence tiers and eras, binned by week. See methodology for details.
Top committees by AI-related spending
Includes party and leadership committees and PACs — large national committees like the RNC and DNC show up among the biggest payers to several vendors here.
5. Who's spending the most — in dollars, and as a share of their budget?
Every candidate committee or PAC/party committee with at least one high-confidence AI-vendor disbursement, across all cycles. "% of total spend" divides AI spend by that committee's total reported operating expenditure that cycle (memo entries excluded to avoid double-counting — see methodology); the share-of-budget table is limited to committees spending at least $5,000 in total that cycle. Click a candidate's name to open their page.
Top by AI spend ($)
Click a column header to sort.
Top by AI spend as a share of total budget
Minimum $5,000 total reported expenditure that cycle. Click a column header to sort.
6. Who's spending on AI on a candidate's behalf?
Everything above is what a candidate's own committee reported paying. This section is different: money spent by Super PACs, hybrid PACs, and party committees — not the candidate's own campaign — to support or oppose a candidate. Independent expenditures (Schedule E) are legally uncoordinated with the candidate; coordinated party expenditures (Schedule F) are a national or state party committee spending on a candidate's behalf, coordinated, up to a statutory per-candidate cap. Both are itemized the same way and matched against the same vendor taxonomy and confidence tiers as the rest of this site — see methodology for sourcing and its limits (notably: money a Super PAC pays a consulting or media-buying firm that itself uses an AI tool is just as invisible here as it is for a candidate's own committee).
Independent expenditures, by vendor
Super PAC / hybrid PAC spending naming an AI vendor, all cycles.
Top independent spenders
"% of spender's IE budget" divides AI-vendor spend by that committee's total reported independent-expenditure spending that cycle. Click a column header to sort.
Every matched independent-expenditure record
All confidence tiers and vendor eras shown, tagged, regardless of the legacy-vendor toggle above — there are few enough records to list individually rather than only as aggregates. Click a column header to sort.
Coordinated party expenditures
Every matched record, all confidence tiers and vendor eras — this schedule's own bulk source has no purpose field, so category labeling is thinner than elsewhere on this site (see methodology). Click a column header to sort.
Methodology & data provenance
This site is generated entirely from public federal disclosure data by an automated pipeline; there is no manual curation of the numbers themselves beyond the vendor/category taxonomy below.
Data sources
Notes & limitations
Full pipeline code, the vendor/category taxonomy (pipeline/config/vendors.yaml), and this site's source are in the GitHub repository.