Research Summary: The New Robber Barons Buy Both Doors

Date: 2026-03-15 Format: Article (~1,500 words) Sources gathered: 14

Topic

The New Robber Barons Buy Both Doors: How AI Super PACs Are Engineering a Congress That Can't Regulate Them

Angle: "Bilateral capture" -- AI companies aren't just lobbying Congress; they're building Congress. By funding candidates in both parties' primaries through parallel PAC structures, the industry ensures that no matter who wins, the regulator is pre-approved.

Thesis Direction

From the mini-brief: The AI industry isn't trying to win elections -- it's trying to make elections irrelevant to AI policy by ensuring both parties' nominees are pre-approved by the industry. This is not lobbying. It is the privatization of candidate selection.

Refined after deep research: The thesis holds and is even stronger than initially suspected. The research reveals a more complete picture:

The mechanism is explicit and documented. Leading the Future operates Think Big (Democrats) and American Mission (Republicans) as formally connected PACs, splitting funds evenly ($5M each). Meta runs Forge the Future (R) and Making Our Tomorrow (D). This isn't coincidence; it's architecture.
There's a direct human link to the proven crypto blueprint. Josh Vlasto -- the same operative who ran Fairshake's Democratic strategy in 2024 -- is now a top official at Leading the Future. The bilateral model is being consciously replicated from crypto to AI.
The policy payoff is already visible. The Trump executive order preempting state AI laws draws directly from an Andreessen Horowitz legal framework. A VC firm's dormant commerce clause argument became federal policy. The Hochul RAISE Act gutting shows the same dynamic at the state level.
The ads-about-everything-except-AI pattern is the tell. Both sides of the AI spending war run ads about immigration, healthcare, and Trump -- never about AI itself. This reveals that the industry knows its actual agenda is unpopular (69-80% of Americans want more AI regulation) and must be advanced through misdirection.
The spending asymmetry is roughly 6:1 anti-regulation. Leading the Future ($125M) + Meta ($65M) vs. Anthropic's Public First ($20M). Even if Public First hits its $50-75M target, the anti-regulation side vastly outspends it.

Suggested thesis refinement for the article:

The AI industry has moved beyond lobbying -- which merely persuades elected officials -- to a new model we should call bilateral capture: the systematic funding of candidates in both parties' primaries to ensure that regardless of which party wins, the winner is pre-vetted by the industry. This is the privatization of candidate selection itself, and it's the most sophisticated threat to democratic accountability since Citizens United.

Evidence Map

Core Structure Evidence (How it works)

Source 02 (NOTUS): FEC filing data showing $5M/$5M split to Think Big/American Mission
Source 03 (Meta $65M): Separate corporate example with Forge the Future (R)/Making Our Tomorrow (D)
Source 01 (NBC News): Both sides' ads avoid mentioning AI entirely

The Blueprint (Where this came from)

Source 05 (Fairshake): Crypto's $290M bilateral model in 2024; Defend American Jobs (R)/Protect Progress (D)/Fairshake (both)
Source 12 (AIPAC model): AIPAC's $100M+ bilateral primary spending preceded both crypto and AI
Source 08 (Popular Info): Josh Vlasto as the direct human link between Fairshake 2024 and Leading the Future 2026

Case Studies (The mechanism in action)

Source 06 (Alex Bores): Former Palantir employee targeted by Palantir co-founder's PAC money for authoring AI regulation
Source 13 (Hochul/RAISE Act): Governor guts strongest state AI law after industry fundraisers
Source 07 (WBEZ Chicago): $62M in four Democratic primaries from converging industry PACs

Policy Outcomes (What the money buys)

Source 10 (Trump EO): a16z legal framework becomes executive order preempting state AI laws
Source 13 (Hochul): Industry-preferred language replaces legislature's bill text

Historical Parallels (This has happened before)

Source 09 (TechPolicy.Press): Big Tobacco/Pharma/Oil tactics mapped to Big Tech
Source 11 (Roosevelt Institute): 163x increase in billionaire spending since Citizens United; measurable policy capture

The Counter-Spending (The losing side)

Source 04 (Anthropic/Public First): $20M vs. $125M+; 69% of Americans want more regulation but the pro-regulation side is outspent 6:1

Internal Tensions (Where the model breaks)

Source 14 (White House pressure): Trump wants partisan loyalty; industry wants bilateral insurance

Strongest Evidence For

The dual-PAC architecture is documented in FEC filings. Leading the Future splits $5M/$5M to Think Big (D) and American Mission (R). Meta runs separate PACs for each party. This is not inference; it is financial architecture designed for bilateral coverage.
The ads avoid mentioning AI. NBC News documented that both pro- and anti-regulation AI PACs run ads about immigration, healthcare, and Trump -- never about AI. This confirms the industry knows its actual agenda is unpopular and must be laundered through other issues.
The crypto playbook was explicitly replicated. The same operative (Josh Vlasto) who ran Fairshake's bipartisan strategy in 2024 is now running Leading the Future. Fairshake spent $290M and proved the model works: Sherrod Brown was defeated, stablecoin legislation passed, industry-friendly regulators were confirmed.
Policy outcomes are already visible. Trump's AI preemption executive order draws directly from an a16z legal framework. Hochul gutted the RAISE Act after industry fundraisers. The spending doesn't just elect candidates; it shapes specific policy.
The spending asymmetry is massive. Anti-regulation: $125M (Leading the Future) + $65M (Meta) + $193M (Fairshake crypto). Pro-regulation: $20M (Anthropic). Public opinion favors regulation 69-80%, but money overwhelms sentiment.

Strongest Evidence Against

This is protected political speech under Citizens United and the First Amendment. The Supreme Court has held that corporations and groups have a right to spend on political advocacy. The AI industry's spending is legally protected, and any attempt to restrict it faces constitutional barriers. Some argue this is democracy working as designed -- industry stakeholders expressing preferences through legal channels.
Industry expertise may improve regulation. A reasonable counterargument: AI is genuinely complex technology, and having industry-informed candidates may produce better regulation than having technologically illiterate lawmakers write rules. The industry argument for a "national framework" (as opposed to 50 state-level regimes) has genuine efficiency logic, even if the industry's preferred version of that framework is self-serving.
Anthropic's counter-spending exists. The pro-regulation side isn't absent -- it's just outspent. Anthropic's $20M donation to Public First shows the AI industry itself is divided, and the "bilateral capture" framing may overstate the uniformity of industry interests.

Research Gaps

Election results are pending. The primaries where AI PACs are spending most heavily (Illinois, New York, Texas, North Carolina) haven't been held yet as of mid-March 2026. We can document the spending and the mechanism, but not yet the electoral outcomes.
Congressional voting records post-capture. We don't yet have data on whether AI PAC-backed candidates who win primaries subsequently vote against AI regulation. The Fairshake/crypto precedent (Randy Fine voting for GENIUS Act) is suggestive but limited.
Internal PAC strategy documents. We know the structure from FEC filings and ads, but we don't have internal strategy memos explaining the bilateral theory in the industry's own words.
Comparative spending efficiency. How effective is primary spending per dollar compared to general election spending or traditional lobbying? The AIPAC data suggests primaries are the highest-ROI vector, but rigorous analysis is thin.
Voter awareness. The Roosevelt Institute notes that voter favorability drops when they learn about super PAC backing. But how aware are primary voters of AI PAC spending? If the ads are "about everything except AI," awareness may be extremely low.

Recommended Approach

Lead with the mechanism, not the outrage. The most distinctive angle for this article is naming the structural innovation: bilateral capture as a concept. Don't just list dollar amounts -- explain why funding both parties' primaries is categorically different from traditional lobbying. The insight is structural, not just financial.

The Bores case study is the human entry point. A former Palantir data scientist who left over the ICE contract, became the country's leading state-level AI regulator, and is now being attacked by his former employer's co-founder using ads about... his work at Palantir. The irony is almost too perfect, and it makes the abstract mechanism concrete.

The crypto blueprint is the proof of concept. Fairshake 2024 should be the "this already worked once" section. Same operative (Vlasto), same structure (parallel partisan PACs), documented policy outcomes (Sherrod Brown defeated, GENIUS Act passed). The AI replication isn't speculative.

The ads-about-everything-except-AI pattern is the most quotable element. The NBC News reporting that AI PACs run ads about immigration and healthcare -- never AI -- is the single most damning piece of evidence that the industry knows its agenda is unpopular and is deliberately hiding it.

Handle the counterargument honestly. The First Amendment/Citizens United defense is legally correct. The industry expertise argument has real force. Don't dismiss these -- engage them and explain why they don't resolve the structural democracy problem.

End with the structural insight, not doom. This article should leave readers with a framework they can reuse: "bilateral capture" as a concept that helps them see this pattern whenever it appears. The path forward isn't obvious, but naming the mechanism is the first step.

Be careful with the AIPAC comparison. The parallel is analytically valid (same bilateral primary spending model), but the AIPAC comparison carries political risk. Use it for structural analysis, not moral equivalence. The point is: AIPAC perfected this mechanism; AI companies adopted it.

Source Inventory

source-01-nbc-ai-ads-about-everything-except-ai.md -- NBC News investigation on AI PAC ads avoiding AI; specific spending data by race
source-02-notus-leading-the-future-70m-war-chest.md -- NOTUS FEC analysis of Leading the Future's financial architecture and donor details
source-03-meta-65m-bipartisan-pacs.md -- Meta's $65M dual-PAC structure (Forge the Future / Making Our Tomorrow)
source-04-anthropic-public-first-20m.md -- Anthropic's $20M counter-spending through Public First Action; pro-regulation strategy
source-05-fairshake-crypto-blueprint.md -- Fairshake's $290M bilateral crypto PAC model from 2024; the proven template
source-06-alex-bores-case-study.md -- Alex Bores as the exemplary target: former Palantir, RAISE Act author, $1.8M+ in attack ads
source-07-wbez-chicago-31m-blitz.md -- WBEZ data on $62M in four Chicago Democratic primaries from converging industry PACs
source-08-popular-info-100m-play.md -- Popular Information on Leading the Future internals, White House pressure, Trump EO connection
source-09-techpolicy-press-tobacco-pharma-parallels.md -- Historical parallels: Big Tobacco/Pharma/Oil regulatory capture tactics adopted by Big Tech
source-10-trump-executive-order-preemption.md -- Trump's AI preemption EO drawing from a16z legal framework; policy payoff of spending
source-11-citizens-united-roosevelt-institute.md -- Roosevelt Institute on 15 years of Citizens United: 163x billionaire spending increase, democracy decline
source-12-aipac-model-comparison.md -- AIPAC's bilateral primary spending model as the structural predecessor to AI PACs
source-13-hochul-raise-act-gutting.md -- Hochul gutting the RAISE Act after industry fundraisers; executive-level capture
source-14-white-house-pressure-bipartisan-tension.md -- White House threatening Silicon Valley over bipartisan PAC spending; internal tensions

Buying Both Doors

Source Material