The fight over AI regulation was always going to get ugly. But nobody expected it to look quite like this: Anthropic and OpenAI, the two companies racing to build the most capable AI systems on earth, are now spending millions on competing super PACs to swing the 2026 midterm elections.
On one side: Public First, backed by Anthropic. On the other: Leading the Future, tied to OpenAI. Both are pouring money into Democratic primaries, trying to elect candidates who'll shape the rules their products will live under. It's not subtle. It's not even trying to be.
The New Normal for AI Companies
This is what happens when an industry matures. First you build the tech. Then you build the business. Then you build the political infrastructure to protect both. We've seen this play out in Big Tech, finance, pharma. AI is just catching up — except it's happening on fast-forward, compressed into months instead of decades.
The New York Times reported that these super PACs are already spending "millions" to influence races. That's real money, and it's coming from companies that, five years ago, were mostly worried about whether they'd run out of GPU capacity.
When your product becomes policy, you stop asking for permission and start buying influence.
There's a certain irony here. Both companies built their brands on safety narratives — Anthropic with its "Constitutional AI" framing, OpenAI with its original "aligned AGI" mission. Now they're doing what every other powerful industry does: making sure the people writing the rules are friendly to their interests.
Why This Matters Beyond the Money
The spending itself isn't the story. The story is what it signals: AI companies no longer see themselves as startups. They see themselves as incumbents. And incumbents don't ask for a seat at the table — they build their own table.
This shift has consequences. When Anthropic and OpenAI become political actors, every policy debate about AI safety, copyright, liability, and export controls becomes harder to separate from their financial interests. The technical arguments don't disappear — but now they're layered on top of campaign donations and PAC spending.
Meanwhile, the economics of AI keep shifting underneath everyone. OpenAI just promoted Greg Brockman to president, making him an "ambassador to the Trump administration" according to the Wall Street Journal. Anthropic is managing secondary market panic as investors try to cash out. These are not the moves of companies that think regulation is someone else's problem.
The Hard Part Comes Next
Here's what I'm watching: this political spending is just the opening bid. The real test is what happens when the candidates these PACs support actually win. Do Anthropic and OpenAI get the regulatory framework they want? Or does the political process do what it always does — produce something neither side expected?
There's also the question of whether this backfires. Voters don't always respond well to obvious attempts to buy influence. And if AI becomes associated with partisan spending in the public mind, that could harden opposition in ways no amount of lobbying can fix.
For now, the money's flowing and the races are heating up. But the companies writing the checks should remember: political capital, unlike compute, doesn't scale linearly. You can't just throw more resources at it and expect the same returns.
The 2026 midterms will tell us whether AI companies learned that lesson — or whether they're about to find out the hard way.