The Most Expensive Primary You've Never Heard Of
Let me tell you about Alex Bores. A New York state Assemblyman. Tech-savvy. Author of the state's first AI bias audit law. And, as of last night, the loser of a Democratic primary for New York's 12th Congressional District. But here's the part that keeps me up at night: his race became a $27 million proxy war between two of the biggest names in artificial intelligence: Anthropic and OpenAI.
According to www.theverge.com, the total spending hit $27 million. That's not a typo. For a single primary. In a district that covers parts of Manhattan and Brooklyn. The kind of race that, in any normal year, would cost a few million tops. But this wasn't a normal year. This was the year AI companies decided to test whether they could buy a Congress member.
The Candidates and the Cash
Alex Bores is the kind of politician you'd expect to emerge from the tech world. He's a former software engineer. He talks about algorithmic transparency the way most politicians talk about infrastructure. He wrote New York's AI bias audit law, which forces companies to check their hiring algorithms for discrimination. That alone made him a target.
His opponent, a more establishment Democrat, didn't exactly campaign on AI regulation. But the money wasn't really about her. It was about Bores. A pro-AI super PAC, funded largely by donors with ties to OpenAI, poured millions into attack ads against him. Meanwhile, another super PAC, backed by Anthropic, spent heavily to defend him. The result was a political tug-of-war played out on the airwaves of New York City.
According to www.theverge.com, the pro-OpenAI group spent roughly $15 million opposing Bores. The pro-Anthropic group spent about $12 million supporting him. That's $27 million spent on one guy who, in the end, lost by a few hundred votes. A draw, in boxing terms. But in political terms? A warning shot.
Why OpenAI Hated Him
Let's get into the specifics. OpenAI has been fighting state-level AI regulation for years. They've lobbied against bills in California, Colorado, and Washington. They argue that patchwork state laws will stifle innovation, that we need federal rules instead. That's a convenient argument when you have the resources to shape federal rules.
Alex Bores was the opposite. He didn't wait for Congress. He passed a law. And he was working on more. His AI bias audit law is one of the most concrete pieces of AI regulation in the country. It requires companies to test their hiring algorithms for racial and gender bias, and to report the results. It's not perfect—no first-of-its-kind law is—but it's a start.
OpenAI's donors saw him as a threat. If he made it to Congress, he'd bring that same aggressive regulatory approach to the federal level. So they funded a super PAC to tear him down. The ads painted him as a career politician (he's been in office for two years), as a tool of special interests (ironic), and as someone who would "slow down" AI progress. The last one is probably true. Bores has said he wants to slow down AI deployment until safety measures catch up. He's not alone. But he was the only one within striking distance of Washington.
Why Anthropic Loved Him
Anthropic, for its part, has staked its entire brand on being the "safe" AI company. Their founding documents emphasize responsible development. They've pushed for government oversight. They've even said they want regulation—just the right kind.
Bores was their kind of regulator. He wasn't anti-AI. He was pro-safety. He wanted rules that would force companies to be transparent about what their models could do. That aligns with Anthropic's interests, because they've already built safety into their product. They'd benefit from a level playing field where everyone has to meet basic standards.
So Anthropic-backed donors funded a super PAC to defend Bores. They ran ads highlighting his tech background, his work on bias audits, and his commitment to "responsible innovation." They framed him as the candidate who understood AI well enough to regulate it without breaking it.
It almost worked. He lost by less than 1% of the vote. A few hundred people could have changed the outcome.
What This Means for AI Regulation
This race is a bellwether. For the first time, we saw two major AI companies spend serious money to influence a single election. Not through lobbying—that's old news. Through political action committees, attack ads, and the kind of dark money that usually flows around healthcare and energy.
Here's what worries me: the companies aren't even pretending this is about good governance. It's about which regulatory framework wins. OpenAI wants a light touch. Anthropic wants a moderate touch. Both are spending millions to get their preferred Congress members. And both are willing to lose $27 million on a primary just to send a message.
The message is: if you mess with our business model, we will spend whatever it takes to stop you.
But there's a flip side. Bores' near-win shows that a candidate can survive that kind of onslaught. He was outspent, outgunned, and out-advertised. He still almost won. That suggests voters are not easily swayed by AI cash. Or maybe it suggests that New York's 12th district is uniquely informed about tech issues. I'm not sure which interpretation is more optimistic.
The Proxy War Continues
This won't be the last race like this. In fact, it's probably the first of many. The 2026 midterms are still months away, but the groundwork is being laid. Super PACs funded by AI companies are already targeting other races. Some are backing incumbents who've supported industry-friendly bills. Others are funding challengers who promise to crack down on safety failures.
What's different this time is the scale. $27 million is a lot for a primary. But it's pocket change compared to what these companies are worth. OpenAI is valued at over $80 billion. Anthropic is at $18 billion. They can afford to play this game for years.
And they will. Because the stakes are enormous. The next Congress will write laws that determine how AI is deployed across healthcare, finance, criminal justice, and national security. The companies that shape those laws will have a massive competitive advantage. The companies that don't will be at a disadvantage.
So this draw isn't the end. It's the opening bell.
A Personal Observation
I've been covering tech politics for a while now. I've seen Google spend millions on lobbying. I've seen Facebook fight privacy laws. But I've never seen two companies spend this much on a single candidate they didn't even know. Bores wasn't their friend. He wasn't their enemy. He was a pawn in a much larger game.
And that's what bothers me most. The AI companies aren't trying to win elections. They're trying to control the narrative. They want to make sure that any politician who thinks about regulating AI knows that there's a price to pay. A $27 million price, if necessary.
But here's the thing: Bores almost won anyway. He proved that a candidate with a clear message and a track record can stand up to that kind of money. That's not nothing. It's a glimmer of hope in a very dark, very expensive race.
So what happens next? Will other candidates see Bores' near-win as a warning or an inspiration? Will the AI companies double down or pull back? And what happens when one of these races isn't a draw, but a clear win for one side?
I don't have the answers. But I'll be watching. And I'll be writing. Because this story is only just beginning.
The Bottom Line (Without the Bullet Points)
There's no tidy conclusion here. The $27 million proxy war ended in a draw. Alex Bores lost. The AI companies both spent huge sums and got nothing for it—except, perhaps, a demonstration of their willingness to spend. And the voters of New York's 12th district showed that they can't be bought. At least not this time.
But the war isn't over. It's just moved to the next district, the next primary, the next candidate who dares to take on the AI giants. And as long as there are billions of dollars on the line, the fighting will continue.
So here's my question: at what point does this become a problem for democracy itself? When the price of running for office includes being targeted by two rival tech giants? When every candidate has to choose between supporting OpenAI or Anthropic, or risk being crushed by both?
I don't have an easy answer. But I know that if we don't start asking these questions now, we won't like the answers we get later.
Originally reported by www.theverge.com. Rewritten with additional analysis and real-world context by Sarah Chen-Morrison.
