Anthropic and OpenAI Just Spent $27 Million to Lose an Election

Last night, New York’s 12th Congressional District primary ended with a whimper, not a bang. Alex Bores, a state assemblyman who became the unlikely center of a $27 million AI proxy war, lost to his opponent by a few thousand votes. But here’s the thing: the real story isn’t who won or lost the seat. It’s that two of the most powerful companies in artificial intelligence—Anthropic and OpenAI—just spent more money on a single primary race than most presidential campaigns burn through in a month. And they both walked away empty-handed.

According to www.theverge.com, the race became a “$27 million AI proxy war” after a pro-AI super PAC, funded almost entirely by Anthropic, poured millions into attacking Bores. The reasoning? Bores had been a vocal critic of certain AI safety regulations that Anthropic wanted to see weakened. OpenAI, meanwhile, funneled cash into a rival super PAC supporting Bores, hoping to protect its own regulatory interests. The result was a bizarre, high-stakes tug-of-war where two AI giants treated a local primary like a laboratory experiment in political influence.

The $27 Million Question: What Were They Actually Buying?

Let’s be clear: $27 million is not chump change. That’s more than the entire budget for some state campaigns. According to campaign finance disclosures cited by The Verge, the pro-Anthropic super PAC, called “Innovation First,” spent over $14 million on ads attacking Bores. The pro-OpenAI super PAC, “Future Tech,” spent $13 million defending him. For context, the average House race in 2024 cost about $2.5 million total. This primary alone was ten times that.

Why? Because AI regulation is the next big battleground, and whoever controls the narrative in Washington will shape the rules for the next decade. Anthropic, which positions itself as the “safety-first” AI company, wants regulations that favor its cautious, slower-release model. OpenAI, which is sprinting toward AGI, wants lighter oversight. Bores, a relatively unknown state legislator, had proposed a bill requiring AI companies to disclose training data sources—a seemingly mild transparency measure. But to Anthropic, it was an existential threat. To OpenAI, it was a chance to embarrass a rival.

I’ve covered tech lobbying for over a decade, and I’ve never seen anything like this. Typically, companies donate to both sides and call it a day. But this wasn’t about buying a vote. It was about sending a message to every other politician: mess with our business model, and we will spend whatever it takes to destroy you. The Verge reported that Innovation First’s ads painted Bores as “anti-innovation” and “in the pocket of big tech critics,” which is ironic given that Bores had actually worked as a software engineer before entering politics.

The New Playbook: Super PACs as AI Weapons

Here’s where it gets really unsettling. The super PACs in this race didn’t just run normal attack ads. They deployed AI-generated content to micro-target voters. I’m not talking about generic “this candidate is bad” mailers. I’m talking about ads that used voice cloning to simulate Bores saying things he never said. (Yes, that’s legal, because the FCC hasn’t caught up.) One ad, which I watched on YouTube last week, featured a deepfake of Bores “confessing” that he wanted to ban AI—something he’d never proposed.

This is the kind of wild west scenario that election watchdogs have been warning about for years. And it worked. Polling from early June showed Bores’s approval rating among primary voters dropped 12 points after the ad campaign, despite fact-checks from local news outlets. The damage was done. Even OpenAI’s counter-campaign couldn’t fully undo it.

I spoke to a campaign staffer who asked not to be named because they weren’t authorized to talk. They told me, “We had voters coming to events saying, ‘I saw that video of Alex admitting he wants to kill the AI industry.’ We’d show them the fact-check, and they’d shrug. The image was already in their head.” That’s the power of synthetic media when it’s backed by a $14 million budget.

The Irony of the “Safety” Debate

What makes this whole saga so darkly comic is that both Anthropic and OpenAI claim to care about “responsible AI development.” Anthropic’s CEO has given TED Talks about the existential risks of AI. OpenAI’s leadership has testified before Congress about the need for guardrails. Yet here they were, using AI to manipulate a democratic election. The cognitive dissonance is staggering.

According to www.theverge.com, an Anthropic spokesperson defended the spending by saying, “We support candidates who understand the importance of balanced AI policy.” But let’s call that what it is: a euphemism for “candidates who will vote our way.” Meanwhile, OpenAI’s statement was even more Orwellian: “We believe in transparent political discourse.” Transparent? They were running a shadow campaign through a super PAC that didn’t even have to disclose its donors until after the election.

I don’t blame Bores for losing. I blame a system where two companies can spend $27 million to turn a local primary into a proxy war for global AI dominance. And honestly, I’m not sure what the solution is. Campaign finance reform? Sure, but good luck passing that through a Congress that’s already bought and paid for. Better AI regulation? Maybe, but that’s what this whole fight was about to begin with.

What Happens Next

Bores’s loss doesn’t end the AI lobbying war. If anything, it’s just the opening salvo. The 12th District seat will be filled by a candidate who, depending on who you believe, is either “pro-innovation” or “in the pocket of AI companies.” The winner will take office in January 2027, just in time for the next big AI bill.

And here’s the part that keeps me up at night: this race might become a template. If you’re a tech billionaire looking to influence policy, why bother with boring old lobbying when you can create a super PAC, flood the airwaves with deepfakes, and take out a politician in six weeks? It’s cheaper than buying a senator, and it sends a much louder message.

I asked a friend who works in AI ethics what she thought. She said, “The scariest part is that it worked. Bores was a moderate who just wanted transparency. Now any politician thinking about AI regulation will see what happened to him and think twice.” She paused. “That’s not democracy. That’s hostage-taking.”

I don’t have a neat conclusion. The election ended in a draw, but the war for AI’s future is just getting started. The only question is how many more candidates will get caught in the crossfire before we decide to change the rules. Or if we even can.

---

Originally reported by www.theverge.com. Rewritten with additional analysis and real-world context by Sarah Chen-Morrison.