Political groups tied to the AI industry spent tens of millions of dollars trying to shape Tuesday’s midterm elections, and the candidates they backed hardest mostly lost. The biggest pro-industry network, Leading the Future, saw its flagship Senate bet collapse in South Carolina, while candidates who campaigned for tougher AI rules won or led in several of the cycle’s most watched races. The results leave the fight over AI regulation — and who gets to write it — wide open.
The industry’s biggest bets lost
Two AI-focused committees, American Mission and Building the Future, spent $4.8 million backing Darline Graham, South Carolina’s Republican Senate candidate. Graham lost to Democrat Annie Andrews, despite the spending, in a state that last sent a Democrat to the Senate in 1998.
American Mission is the Republican arm of Leading the Future, a network backed by OpenAI president Greg Brockman and venture firm Andreessen Horowitz. It spent more than $500,000 each on Republican House candidates Aaron Flint in Montana and Laurie Buckhout in North Carolina. Both lost, while the network’s Democratic arm, Think Big, backed Florida Representative Jared Moskowitz, who won reelection.
The other side spent too
Groups pushing for stronger AI safeguards had mixed results of their own. Public First, backed by Anthropic, supported several Republican candidates who lost, but did better with Democrats, including a $3 million contribution to a committee backing California state Senator Scott Wiener, who was leading his race when the results were tallied.
The Guardrails Alliance, which campaigns to make AI-industry money a political liability, spent more than $1 million each opposing California Representative Jimmy Gomez and Montana’s Aaron Flint. It also spent $300,000 backing Angela Gonzales-Torres, who beat Gomez. Elon Musk, who reportedly poured about half a billion dollars into backing Republicans, also saw many of his candidates lose.
The wider count is worse for the industry
An independent tally by the newsletter Citation Needed puts crypto and AI spending at $250 million combined, and the record in the races that mattered is stark. Of the twelve competitive races called so far, the side backed by crypto and AI super PACs lost nine.
Those groups spent $74.7 million on the losing side of those races, against $28.3 million on the winning side. The pattern held across the map: in Ohio, Democrat Sherrod Brown beat Jon Husted despite more than $64 million in crypto money spent against him, the most the industry has ever put into a single race.
In Michigan, Democrat Abdul El-Sayed won the Senate seat despite more than $1 million from the conservative AI group Alliance for a Better Future backing his opponent. In Maine, the same group spent $810,000 boosting Susan Collins, who lost her seat. In Colorado, Manny Rutinel unseated Gabe Evans, who faced $600,000 in opposition spending from the Public First network.
Our read: safe seats hid the real record
The headline for the industry is not that it lost, but where it lost. The same tally shows a safe-seat strategy: in forty safe races, their picks went forty for forty, while in the twelve competitive races, they lost nine. Money still buys influence in primaries — Leading the Future spent $27.38 million there — but not competitive general elections.
The central policy fight is now sharper, not settled. Both camps agree the question is whether Congress writes one national AI framework that overrides state laws, or states keep their own safeguards. With Democrats taking both chambers, according to Chosun Biz, and regulation advocates winning seats, the industry’s single framework faces a harder path — while states keep legislating anyway.
The same skepticism showed up beyond the ballot for candidates. Ohio voters backed local bans on AI data centers, and the election redrew the industry’s buildout map, with several states and towns now weighing limits on new facilities and power deals.
Why It Matters
This was the AI industry’s first electoral test, and it failed where it counted. A Congress more open to regulation, and states already writing their own rules, puts the industry’s bet on a single, industry-friendly national framework in doubt. The groups say they will keep spending through 2028, so the next test comes in the primaries, not the general election.
Background
AI companies spent the election season preparing for the vote itself: OpenAI set up an election operations center, and tools like Microsoft’s Copilot were tested on voting questions, getting one in four wrong. The midterms were the first national vote since those tools became part of everyday life.
Source: TechCrunch’s midterm tally and Citation Needed’s election results
0 Comments