Article Thesis

Working Title

Buying Both Doors

Subtitle

The AI industry isn't lobbying Congress. It's building Congress -- funding candidates in both parties' primaries so that no matter who wins, the regulator was pre-approved by the regulated.

Thesis

The AI industry has invented a form of political influence that makes traditional lobbying look quaint: bilateral capture, the systematic funding of candidates in both parties' primaries to ensure that regardless of which party wins a given seat, the winner is pre-vetted by the industry. This isn't corruption in the classic sense -- no one is being bribed. It's something structurally worse: the privatization of candidate selection itself. When the industry funds both doors, the election still happens, the voters still vote, but the menu of choices has already been curated by the people who need to avoid regulation. Democracy's front end still works. The back end has been captured.

The Framework

Bilateral capture is the conceptual engine. Traditional lobbying persuades legislators after they're elected. Bilateral capture selects them before they're nominated. The mechanism works in three stages:

Architecture. Establish parallel PACs -- one for each party's primaries. Leading the Future runs Think Big (Democrats) and American Mission (Republicans), splitting funds evenly. Meta runs Forge the Future (R) and Making Our Tomorrow (D). The structure itself is the strategy: regardless of partisan outcome, the winner owes their margin to industry money.
Misdirection. Run ads about everything except the policy you're buying. AI PAC ads are about immigration, healthcare, and Trump -- never AI. This is the tell. It confirms the industry knows its actual agenda (blocking regulation that 69-80% of Americans support) is unpopular, so the real transaction must be laundered through other issues.
Harvest. Once the candidates are seated, collect the policy return. A venture capital firm's dormant commerce clause argument becomes a presidential executive order preempting state AI laws. A governor rewrites the strongest state AI bill to match the industry's preferred weaker standard after accepting industry fundraiser money. The investment matures.

What makes bilateral capture different from ordinary political spending is that it doesn't try to win elections -- it tries to make elections irrelevant to the policy question. It's the difference between playing the game and rigging the bracket. The term should make people see this pattern everywhere it appears -- not just AI, but crypto (Fairshake's $290M playbook), pro-Israel politics (AIPAC's $100M+ primary model), and whatever industry adopts it next. Once you see bilateral capture, you can't unsee it.

Why This Matters Now

Three forces are converging right now. First, the AI industry's $190M+ in PAC money is flowing into the first primaries of the 2026 cycle -- Texas, North Carolina, Illinois -- and the structural pattern is fully visible in FEC filings for anyone who looks. Second, the crypto industry already proved this model works: Fairshake's 2024 bilateral spending defeated Sherrod Brown, passed the GENIUS Act, and installed industry-friendly regulators, and the same operative (Josh Vlasto) who ran that playbook is now running the AI version. Third, Trump's December 2025 executive order preempting state AI laws -- built on an Andreessen Horowitz legal framework -- shows the policy pipeline is already delivering returns before the 2026 elections are even held. The window to name this pattern and make it politically costly is closing. Once the primaries are over and the captured candidates are seated, the mechanism becomes invisible.

The Hook

Open with the ads themselves. A voter in a Chicago-area congressional district sees an attack ad about ICE. Another about healthcare costs. Another about Trump. None of them mention AI. But all of them were paid for by AI companies. $62 million is flooding four Democratic primaries in Illinois alone -- from AI, crypto, and AIPAC PACs -- and the voters being targeted have no idea what they're actually being sold. The ads are about everything except the thing the money is buying. Start there. That gap between what the voter sees and what the money wants is the entire article in miniature.

Key Evidence & Sources

FEC filings showing the dual-PAC architecture: Leading the Future transferred $5M to Think Big (D) and $5M to American Mission (R) -- a perfectly balanced bipartisan split documented in federal records. (NOTUS)
The ads-about-everything-except-AI pattern: NBC News documented that competing AI PACs run ads about ICE, immigration, healthcare, and Trump -- never about AI itself. Both sides do this, confirming the misdirection is structural, not accidental.
The Vlasto bridge: Josh Vlasto, the Democratic operative who ran Fairshake's 2024 crypto strategy (which defeated Sherrod Brown and passed industry-friendly legislation), is now a top official at Leading the Future. Same person, same playbook, new industry.
The Bores case study: Alex Bores, a former Palantir data scientist who left over the ICE contract and became the country's leading state-level AI regulator, is being attacked by a PAC funded by Palantir co-founder Joe Lonsdale -- using ads about Bores's work at Palantir. $1.8M+ in negative ads from the very industry he tried to regulate.
The Hochul RAISE Act gutting: Governor Hochul crossed out the entire text of the state's strongest AI safety law and replaced it with industry-preferred language after Ron Conway (a Leading the Future donor) held fundraisers for her. Dual strategy: gut the bill AND punish its author through electoral spending.
Trump's AI preemption executive order: The legal backbone of the December 2025 EO preempting state AI laws came directly from an Andreessen Horowitz dormant commerce clause argument published three months earlier. A VC firm's legal theory became federal policy.
The spending asymmetry: Anti-regulation PACs ($125M Leading the Future + $65M Meta) outspend the pro-regulation side (Anthropic's $20M Public First) by roughly 6:1 -- despite 69-80% of Americans wanting more AI regulation.
Citizens United as infrastructure: Billionaire election spending has increased 163x since 2010. Over 80% of 2024 billionaire spending used channels that were prohibited before Citizens United. Bilateral capture is the most sophisticated expression of this 15-year trend. (Roosevelt Institute)

Argument Arc (Brief)

Open with the gap -- the Chicago voter who sees attack ads about everything except the thing the money is actually buying. This is the experiential entry point: something is being done to voters that they can't see. Name the mechanism -- introduce bilateral capture as a concept, using the FEC-documented dual-PAC architecture (Leading the Future's Think Big/American Mission, Meta's Forge the Future/Making Our Tomorrow) to show this isn't inference but documented financial engineering. Show the blueprint -- Fairshake's 2024 crypto playbook proved the model works, and the same operative carried it to AI. This isn't theoretical; it already succeeded once. Make it human -- the Bores case study, where a former Palantir employee is attacked by his former employer's co-founder for having worked at Palantir, puts a face on the abstraction. The Hochul RAISE Act gutting shows the policy return in action. Steelman honestly -- the First Amendment defense is legally correct, industry expertise has real value, and Anthropic's counter-spending shows the industry isn't monolithic. Engage these, then explain why they don't resolve the structural democracy problem: legal doesn't mean democratic, expertise doesn't require capture, and a 6:1 spending ratio isn't a real debate. Land on the structural insight -- bilateral capture is a reusable concept. The reader should walk away seeing this pattern wherever it appears and understanding that the threat isn't that elections are being stolen, but that they're being pre-decided at the primary stage where nobody's watching.

The "So What?"

The reader should walk away understanding that the AI industry's political spending represents something categorically different from traditional lobbying -- not persuasion of elected officials, but curation of the candidate pool itself. The concept of bilateral capture should become a lens they apply going forward: whenever they see industry money flowing to both parties' primaries with ads about everything except the industry's policy agenda, they'll recognize the pattern. The deeper insight is about where democratic accountability actually lives. We assume it lives in the general election, but bilateral capture moves the decisive moment upstream to the primary, where turnout is low, voter awareness is minimal, and a few million dollars can determine who appears on the ballot. The general election becomes a formality -- a choice between two pre-approved options. Once you see that the real election happened in the primary, and the real primary was decided by money, the whole system looks different.

Potential Pitfalls

The AIPAC comparison is analytically valid but politically charged. The structural parallel (bilateral primary spending, ads about everything except the policy agenda) is real and important. But the AIPAC comparison triggers a separate and intense political debate that could overwhelm the article's argument. Use it precisely and structurally -- AIPAC perfected the mechanism, AI companies adopted it -- without implying moral equivalence or getting drawn into the Israel-Palestine debate.
"Bilateral capture" could sound like just a new name for something old. Industries have always spent on both parties. The article needs to clearly distinguish what's new: the targeting of primaries rather than generals, the dual-PAC architecture designed for bipartisan coverage from inception, and the misdirection strategy of running ads about unrelated issues. It's the precision engineering that's novel, not the general principle.
The elections haven't happened yet. We can document the spending and the mechanism, but not the electoral results. The article should be honest about this: we're naming the strategy in real time, before the outcome is known. The Fairshake/crypto precedent is the best evidence that the strategy works, but the AI cycle is still in progress.
The Anthropic counter-spending complicates the "unified industry" framing. Anthropic is an AI company spending $20M on the pro-regulation side. The article shouldn't pretend the industry speaks with one voice. But the 6:1 spending ratio tells the real story: the pro-regulation faction exists but is massively outgunned.
Legal vs. democratic. The strongest counterargument is that this spending is constitutionally protected. The article must engage this honestly -- yes, it's legal under Citizens United -- while making the case that legal and democratic are not the same thing. The Constitution protects the speech; it doesn't require us to pretend the speech isn't corrosive.

Research Assessment

The source material is exceptionally strong -- 14 substantive sources covering every facet of the argument: the financial architecture (FEC filings), the misdirection strategy (NBC News ad analysis), the proven blueprint (Fairshake/crypto), the human case study (Bores), the policy payoff (Trump EO, Hochul RAISE Act), historical parallels (tobacco/pharma/oil), structural context (Citizens United/Roosevelt Institute), the AIPAC model, and the internal tensions (White House pressure). No significant supplemental research was needed. The one area that could use deeper sourcing during the draft stage is comparative data on primary vs. general election spending efficiency -- the intuition that primaries are the highest-ROI vector for outside money is well-supported by the AIPAC data but could benefit from political science research quantifying the turnout-to-spending ratio.