An AI Donor Cluster in Federal Campaign Finance

Community detection on FEC contribution data

political science
networks
campaign finance
Author

Jesse Brandt

Published

June 23, 2026

I analyzed data from the United States Federal Election Commission (FEC) on individual contributions for the 2026 election cycle. I limited the scope of the analysis to large contributions (>$3,000) to candidate committees for those running for House, Senate, or President.

The goal of this exploratory analysis is to identify patterns and communities among donors. For individual contributions, since contribution limits cap how much one donor can give, for candidates to raise large amounts of cash, they need to court many donors. This may include ‘bundlers,’ who coordinate many individual donors to funnel large amounts of money to their preferred candidates.

To find patterns and potentially detect bundling, I structured the data as a network. Each donation over $3,000 is a link between a donor and a candidate, and I projected these links into donor-donor connections: if two donors give over $3,000 to the same candidate 3 or more times, they are linked, and these links are weighted by the number of shared candidates.

The cluster

Louvain community detection (Blondel et al. 2008) on a donor-donor projection of FEC contribution data (edges where two donors share 3+ candidate recipients) identified 77 communities. One community has a notably high concentration of AI industry employees.

Force-directed network graph showing six donor communities detected by Louvain algorithm. Each community forms a visible cluster. The AI cluster, shown in orange, is a smaller but densely connected group. Node size reflects degree.

Where the money goes

The top recipient of donations from this cluster is Alex Bores. Bores has drawn attention for the AI-related PACs spending both for and against him (Wilkins 2026). This analysis reveals financial support from individuals in the AI industry as well.

All amounts from contributions over $3,000.

Candidate Donors Contributions Total
ALEX BORES FOR NY 77 154 $539,000
SCOTT WIENER FOR CONGRESS 72 142 $506,700
RO FOR CONGRESS INC 72 135 $483,500
RAJA FOR ILLINOIS 71 111 $387,500
MANNY RUTINEL FOR CONGRESS 70 121 $429,800
ERIC JONES FOR CONGRESS 31 59 $563,633
FRIENDS OF RAJA FOR CONGRESS 26 49 $167,800
WAHLS FOR IOWA 19 35 $122,400
TOM PERRIELLO FOR CONGRESS 15 28 $98,000
TED LIEU FOR CONGRESS 12 25 $94,500
LICCARDO FOR CONGRESS 11 17 $58,700
RAKHI ISRANI FOR CONGRESS 9 16 $56,000
HICKENLOOPER FOR COLORADO 9 15 $52,500
FRIENDS FOR GREGORY MEEKS 7 12 $41,800
NEIL FOR CONGRESS 7 7 $24,500

Most connected donors

Degree = number of other cluster members who share at least 3 candidate recipients. Contributions and totals from contributions over $3,000.

Donor Employer Degree Betweenness Contributions Total
BILLS, STEVEN ANTHROPIC 82 3479.76 43 $154,000
LIN, TAO ANTHROPIC 82 983.43 30 $111,856
ZIEGLER, DANIEL ANTHROPIC 77 866.80 35 $128,000
FOSTER, LAUREN AGE BOLD 72 66.98 21 $73,500
THOMAS, DRAKE ANTHROPIC 72 310.85 17 $59,500
LOFGREN, PETER ANTHROPIC 71 81.12 18 $71,100
SASTRY, GIRISH GIRISH SASTRY 71 77.87 21 $80,500
BRINICH-LANGLOIS, PATRICK ACADEMIA.EDU 70 137.15 13 $44,900
MAVRIDES, DYLAN JANE STREET 69 29.63 11 $38,100
KIJEWSKI, JOSEPH EVERLAW 67 20.23 13 $45,500
BORGESON, BLAKE SELF EMPLOYED 65 39.40 12 $42,000
FARHI, DAVID SAFE SUPERINTELLIGENCE 63 11.79 13 $45,500
GILBERT, JESSE CIVAI 63 34.24 9 $31,500
LEVIN, TREVOR OPEN PHILANTHROPY 63 5.79 6 $21,000
ABELE, EMMA PROTECTIVE EQUIPMENT INC. 62 10.81 12 $49,000
KELLERMANN, ANTONY SELF EMPLOYED 62 10.81 11 $38,500
MARKOV, TODOR OPENAI 62 11.50 25 $90,600
MENNEN, ALEX PURE STORAGE 62 10.81 10 $34,900
NEYMAN, ERIC ARC 62 10.81 12 $47,800
SMITH, MICHAEL CORNERSTONE RESEARCH 62 316.00 32 $122,100

Who’s in it

Top employers by number of donations (contribution-level, raw self-reported strings).
Employer Donations Donors Total
ANTHROPIC 124 13 $448,700
NOT EMPLOYED 124 36 $439,800
SELF EMPLOYED 55 23 $193,900
ANTHROPIC PBC 54 6 $194,356
OPENAI 38 7 $143,000
(blank) 33 28 $127,500
GOOGLE 30 5 $105,000
OPEN PHILANTHROPY 29 6 $102,800
SELF-EMPLOYED 28 11 $110,600
N/A 22 14 $77,000
SELF 21 9 $77,000
AI LAB WATCH 19 1 $73,500
WANXIANG AMERICA CORPORATION 16 1 $59,500
KAITAR RESOURCES 15 1 $63,000
AVERI 15 1 $51,443
ELECTRIC CAPITAL 14 1 $49,000
SAFE SUPERINTELLIGENCE 13 1 $45,500
EVERLAW 12 1 $42,000
XYZ VENTURE CAPITAL 12 1 $42,000
ALIGNMENT RESEARCH CENTER 11 1 $41,200
CENTRE FOR EFFECTIVE ALTRUISM 11 2 $38,500
NESSEL DEVELOPMENT 11 1 $37,950
CONFUSION CAPITAL 10 1 $38,500
GIRISH SASTRY 10 1 $35,000
PURE STORAGE 10 1 $34,900

The network

Network graph of 157 donors in this cluster. Orange nodes have employers on an ad-hoc AI-related list; grey nodes do not. The most connected nodes are labeled.

Cluster subgraph. Node size = degree. Orange = employer on ad-hoc AI-related list.

Geography

State Donors
CA 95
IL 18
NY 11
MA 7
TX 6
DC 5
FL 3
WA 2
AZ 1
CO 1
CT 1
DE 1
GA 1
MO 1
NM 1
NV 1
PA 1
VA 1

This cluster is mostly California-based, despite Bores’ location in New York.

Timing

The Bores donations from this cluster mostly occur in October 2025. This suggests a shared impetus for the donations rather than just an ideological alignment.

Caveats

  • Donor identity is name-only. Two people with the same name in different states are conflated; the same person with different spellings is split.
  • Louvain resolution is fixed at 1.0 (the default). Community structure is sensitive to this parameter. These communities have not been tested for stability across resolutions.
  • Structural patterns, not proof of coordination. Community detection identifies donors with similar giving patterns. It does not establish that giving was coordinated.
  • All dollar amounts are from individual contributions over $3,000 to candidate-authorized committees (FEC types H, S, P). The federal per-election contribution limit is $3,300 (2023–2024) / $3,500 (2025–2026); the $3,000 threshold is approximate.

References

Bates, Douglas, Martin Maechler, and Mikael Jagan. 2026. Matrix: Sparse and Dense Matrix Classes and Methods. https://doi.org/10.32614/CRAN.package.Matrix.
Blondel, Vincent D., Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. 2008. “Fast Unfolding of Communities in Large Networks.” Journal of Statistical Mechanics: Theory and Experiment 2008 (10): P10008. https://doi.org/10.1088/1742-5468/2008/10/P10008.
Csárdi, Gábor, Tamás Nepusz, Vincent Traag, et al. 2026. Igraph: Network Analysis and Visualization. https://r.igraph.org/.
Mühleisen, Hannes, and Mark Raasveldt. 2026. Duckdb: DBI Package for the DuckDB Database Management System. https://r.duckdb.org/.
Pedersen, Thomas Lin. 2024. Tidygraph: A Tidy API for Graph Manipulation. https://tidygraph.data-imaginist.com.
Pedersen, Thomas Lin. 2025. Ggraph: An Implementation of Grammar of Graphics for Graphs and Networks. https://ggraph.data-imaginist.com.
R Core Team. 2026. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing. https://doi.org/10.32614/R.manuals.
Wilkins, Emily. 2026. “Dueling PACs Take Center Stage in Midterm Elections over AI Regulation.” CNBC, February 19. https://www.cnbc.com/2026/02/19/dueling-pacs-take-center-stage-in-midterm-elections-over-ai-regulation.html.

Data: FEC individual contributions (bulk files). Analysis conducted in R with igraph, tidygraph, and ggraph.