AI companies are spending up to $185 million to shape the 2026 midterms through super PACs whose ads never mention AI. The money mostly loses — except when nobody is watching.
NBC News ran the headline in March: Ads funded by AI industry are flooding the 2026 election. They're about everything except AI.
The ads attack candidates on crime, immigration, housing. They praise candidates for supporting veterans, small businesses, public schools. The organizations funding them have names like Leading the Future, Public First Action, and the American Technology Excellence Project. Nothing in the names, the ads, or the issue positions tells you that AI companies are writing the checks.
The Money
Leading the Future has raised a hundred and twenty-five million dollars for the 2026 midterms. Its co-founders are Greg Brockman, the co-founder of OpenAI, and Joe Lonsdale, the co-founder of Palantir. Marc Andreessen and Ben Horowitz of a16z contributed roughly fifty million. The PAC is explicitly modeled on Fairshake — the crypto industry super PAC that spent a hundred and ninety-one million dollars on the 2024 cycle — and shares the same strategist, Josh Vlasto.
Anthropic contributed twenty million dollars to Public First Action, a separate super PAC that has raised fifty million and expects to reach seventy-five. Meta launched its own California-focused PAC with twenty million and a multi-state PAC with forty-five million more. Combined with Leading the Future and smaller efforts, AI industry super PAC spending on the 2026 midterms is tracking between a hundred and twenty-five and a hundred and eighty-five million dollars.
The architecture is borrowed from crypto. The playbook is borrowed from crypto. The only difference is that the crypto PACs told voters they were about crypto. The AI PACs don't mention AI at all.
The Fight Inside the Fight
OpenAI and Anthropic are on opposite sides. Leading the Future — backed by OpenAI's co-founder — and Public First Action — backed by Anthropic — are funding rival candidates in the same primaries. In New York's twelfth congressional district, Think Big PAC ran attack ads against Alex Bores, a state assemblymember and AI safety advocate running for Congress, tying him to donations from Sam Bankman-Fried's 2022 political network. Bores's campaign sent a cease-and-desist letter calling the ads false and defamatory.
The two companies agree on one thing: the ads should be about anything other than AI. Both PACs run on hot-button issues. Neither mentions the regulatory stakes that motivated the spending.
The real fight is over the form of regulation — who writes the rules, which agency enforces them, whether state or federal law governs. Industry addresses the form while concealing the substance from voters. An NBC poll in March found that fifty-seven percent of voters believe the risks of AI outweigh the benefits. Both sides of the AI PAC war apparently concluded that running on AI policy would lose.
The Immune System
Here is the part the money didn't expect. In Illinois, the spending mostly failed.
Fairshake spent roughly ten million dollars opposing Stratton in the Illinois primary. Stratton won. Think Big went one-for-two — it won with Melissa Bean in the eighth district but lost in the second. Bloomberg's headline after the results: AI and crypto suffer surprise defeats. When voters paid attention to the races, the dark money didn't convert.
The mechanism is attention. High-profile primaries concentrate voter scrutiny. Spending is visible. Endorsements are tracked. Attack ads are parsed. When the audience is paying attention, the ventriloquist's lips are visible.
The Low-Attention Race
But not every race gets attention.
In North Carolina's fourth congressional district, Jobs and Democracy PAC — an Anthropic-backed entity operating under the Public First umbrella — spent roughly one point six million dollars supporting Valerie Foushee. She won by one thousand two hundred and two votes. A margin that small means the PAC spending was almost certainly decisive.
The pattern: AI super PAC money fails in races where voters are watching, and succeeds in races where they aren't. This is not a bug in the strategy. It is the strategy. You don't need to win the races everyone follows. You need to win the ones nobody follows. The fewer eyes on a district, the more a million dollars buys.
The Double Invisibility
The first invisibility is the money itself. Voters in North Carolina's fourth district may never learn that an AI company's PAC helped choose their representative. The ads were about local issues. The funding source was a name that means nothing to someone not tracking FEC filings.
The second invisibility runs deeper. Even if voters detect the money — even if every PAC dollar were disclosed in real time — they cannot detect what the money is optimizing for. The AI industry is not buying votes on a specific bill. It is buying a cohort of legislators pre-selected for compatibility with industry priorities. The PAC strategists are not choosing candidates who will vote yes on a particular regulation. They are choosing candidates who will never write one.
When the ventriloquist is skilled enough, the audience stops looking for lips. They watch the puppet and believe it speaks for itself. A hundred and eighty-five million dollars is a bet that American voters won't look up.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)