The Midterm Ads Never Mention AI. The Checks Come from OpenAI.

Introduction

A New York assemblyman who wrote an AI safety bill just got hit with $1.8 million in attack ads about his former employer's ICE contracts. In North Carolina, a congresswoman received $1.6 million in support ads about fighting immigration raids. The checks clearing those buys came from OpenAI's co-founder and Marc Andreessen.

The $300M Silence

The AI industry has committed over $300 million to the 2026 midterm elections through a network of super PACs. That number is a mix: $50.3 million verified through FEC filings for Leading the Future, $100 million pledged by the new Innovation Council Action, $65 million committed by Meta across state-level vehicles, and roughly $50 million raised by the pro-regulation side (mostly Anthropic's $20 million). The spending ratio between pro-deregulation and pro-regulation groups lands around 5-to-1.

That scale alone would be a story. But the weirder part is what the ads actually say. NBC News documented the pattern across New York, Illinois, North Carolina, and Texas: AI-funded PACs are running ads about immigration enforcement, healthcare, the Affordable Care Act, Trump accountability, and generic biographical spots. Almost none mention artificial intelligence. Brad Carson, co-leader of Public First Action (the Anthropic-backed, pro-regulation PAC), told NBC: "we know AI isn't the first thing on every voter's mind when they go to the polls." Even the side pushing for AI safety rules won't put the letters A-I in its own ads.

This is the crypto PAC playbook from 2024, scaled up. FairShake ran the same structure in that cycle: bipartisan super PAC arms, massive war chest, wedge-issue ads that never mentioned cryptocurrency. It worked well enough that AI donors noticed. Leading the Future hired Josh Vlasto, who previously advised FairShake, to run the operation.

The Bores Test Case

Alex Bores is a New York State Assembly member running for Congress in NY-12. He authored the RAISE Act, which would require AI companies to disclose safety protocols and report serious misuse incidents. That bill made him a target.

Leading the Future's Democratic arm, Think Big, spent $1.8 million and counting on attack ads against Bores. The ads focused on his former employer Palantir's contracts with ICE, framing him as complicit in immigration enforcement. Think Big also ran ads linking him to SBF's political donations. The ads ran on immigration and crypto fraud — not the bill he actually wrote.

The Anthropic-backed Jobs and Democracy PAC countered with $450K in ads defending Bores. So a state lawmaker who tried to regulate AI companies is now the subject of a $2.25 million proxy fight between two AI-funded super PACs, and voters in his district are watching ads about ICE raids and Sam Bankman-Fried.

$1.8 million to attack, $450K to defend, over a bill that would have required AI companies to disclose safety protocols.

Follow the Person: Sacks to Scorecard

David Sacks served as White House AI and crypto czar from inauguration through roughly March 26, 2026. During his time in government, he shaped the executive order targeting state AI laws and the March 20 national AI legislative framework calling for Congress to enact a single federal statute preempting state AI regulation. He maintained investments in 400+ tech firms with AI ties through his fund Craft Ventures, operating on ethics waivers that Kathleen Clark, a government ethics expert, called "sham ethics waivers."

The policy Sacks helped write has a specific goal: federal preemption of state AI laws. Congress has rejected that approach twice already, once in the One Big Beautiful Bill Act and once in the NDAA. The legislative path stalled. So three days after leaving government, on March 29, Sacks endorsed Innovation Council Action, the $100 million PAC that will spend the midterms enforcing the exact deregulation agenda he wrote from inside the White House. If Congress won't preempt state laws by voting for it, the industry can change who's in Congress. He moved to co-chair PCAST (the President's Council of Advisors on Science and Technology), keeping policy influence without the ethics disclosure requirements that come with his former role.

The PAC itself is run by Taylor Budowich, who served as Trump's deputy chief of staff for communications until September 2025, six months before the PAC's announcement. The revolving door here has different speeds: Sacks went government-to-endorsement in three days, Budowich took six months — same destination, different commutes.

Who Benefits

The direct beneficiaries are the companies writing the checks. Greg Brockman (OpenAI co-founder) and his spouse gave $25 million combined. Marc Andreessen and Ben Horowitz gave $25 million combined. Meta committed $65 million, its largest-ever election investment. The policy they're buying is federal preemption of state AI laws. Right now, 27 states have introduced 78 AI chatbot safety bills, part of a broader wave of nearly 1,100 AI bills tracked across all 50 states this session. A single light federal framework, written with industry input, replaces all of that. The financial incentive is compliance cost reduction at massive scale.

Sacks benefits personally and directly. His Craft Ventures portfolio holds 400+ AI-adjacent investments. A friendlier regulatory environment raises the value of every one of them. NPR documented one pipeline in detail: Craft Ventures invested in BitGo, Sacks backed the GENIUS Act (stablecoin legislation) while in office, BitGo celebrated the act and filed for an IPO with Craft owning 7.8%.

Trump's political operation benefits without spending a dollar. Innovation Council Action grades lawmakers on loyalty to the president's AI position, and AI industry money flows to Trump-aligned candidates as a result. The president's agenda gets an enforcement mechanism funded entirely by Silicon Valley, one that costs his campaign nothing and disciplines members of his own party on tech votes.

The Scorecard That Doesn't Exist Yet

Here's the part that jumped out to me. Innovation Council Action announced it would build a scorecard ranking every member of Congress on support for AI deregulation. Criteria include backing a single federal regulatory framework, faster data center approvals, fewer barriers to AI development, and opposition to state-level safety mandates. The scores guide where the money goes: donations, endorsements, attack spending, and primary challenges.

But the scorecard hasn't been published. No methodology, no grades, no public criteria, no timeline for release beyond what Axios reported. And that might be the whole point. The threat of being scored, with $100 million behind the grading curve, changes behavior before any score exists. A lawmaker considering an AI safety vote doesn't need to see their grade. They need to know the grading system is coming.

To be fair, the ads-about-everything-except-the-real-issue approach is not unique to AI money. AIPAC, environmental groups, and labor unions all do this. It's legal, it's common, and it works. The counterargument is that voters should be influenced by issues voters care about, not issues donors care about. But that's a structural critique of super PACs as a category, not something specific to tech.

What's specific to tech is the scale and the coordination with active White House policy. AI-backed candidates have advanced in 10 of 11 congressional primaries through late March 2026, though several of those were advances to runoffs rather than outright wins. Illinois showed the limits: Think Big spent $1.4 million on Jesse Jackson Jr. in IL-2 (he lost) and $1.1 million on Melissa Bean in IL-8 (she won). The results prompted a blunt summary from reporters covering the primaries: "AI and Crypto spent nearly $20 million to reshape Illinois's Democratic primaries. They mostly lost." Winning individual races isn't really the point. The goal is building a system where lawmakers track their AI score the same way they track their NRA rating — before the money even moves.

What Comes Next

Q1 2026 FEC filings drop April 15. That's when we'll see whether Leading the Future's reported $125 million matches what's actually been filed, and where Meta's $65 million landed race by race. The scorecard still hasn't been published. When it does come out, check whose names are on it and which of those members have AI bills pending in committee.