Everyone loves to pretend AI companies are above the dirty business of politics. They’re not.
Anthropic just filed paperwork for AnthroPAC, a shiny new political action committee that’ll funnel employee donations to candidates in the 2026 midterms. Both parties, naturally. Can’t play favorites when you’re trying to keep regulators off your back.
This is the same company that positioned itself as the “safety-first” alternative to OpenAI. The responsible adults in the room. The ones who care about alignment and ethics and all those warm fuzzy concepts that make venture capitalists feel better about dumping billions into glorified autocomplete engines.
Follow the Money
Let’s be clear about what’s happening here. Anthropic already dropped $20 million on Public First Action back in February—a group focused on AI safeguards. That’s the PR-friendly donation. The one you announce proudly. The one that says “we’re the good guys.”
AnthroPAC is different. This is about access. About making sure the right people pick up the phone when Anthropic calls. About ensuring that whatever regulations eventually pass don’t hurt too much.
The structure is clever, I’ll give them that. Employees donate up to $5,000 per candidate, not the company directly. It’s cleaner this way. More democratic-looking. But let’s not kid ourselves about who benefits when Anthropic’s workforce starts writing checks to lawmakers.
Why Now?
Timing matters. We’re heading into midterms with AI regulation actually on the table. Not the vague “we should probably do something” discussions of 2023, but actual proposed legislation. Export controls. Safety requirements. Liability frameworks.
Every major AI company is scrambling to shape these rules before they get shaped for them. Meta, Google, OpenAI—they’ve all been ramping up their Washington presence. Anthropic was late to this party, and they know it.
The company’s recent legal battle with the Pentagon probably didn’t help their comfort level either. When you’re fighting the Department of Defense in court, having some friends on Capitol Hill starts looking pretty attractive.
The Safety Theater Problem
Here’s what bugs me most about this move. Anthropic built its entire brand on being different. On caring more about safety than speed. On having a “constitution” for Claude that supposedly keeps it aligned with human values.
But forming a PAC to donate to politicians? That’s the oldest corporate playbook in existence. There’s nothing principled about it. Nothing that suggests deep thinking about AI’s societal impact.
It’s just business. Expensive, necessary business if you want to survive in a regulated industry. But business nonetheless.
What This Means for You
If you’re using Claude, this doesn’t change anything about the product. The chatbot will work the same tomorrow as it did yesterday.
But if you believed Anthropic was somehow fundamentally different from other AI companies—more ethical, more careful, more aligned with public good over private profit—well, adjust your expectations.
They’re playing the same game as everyone else. They’re just better at the messaging.
AnthroPAC will make its donations. Candidates will cash the checks. Some will win, some will lose. And when the dust settles, Anthropic will have relationships with lawmakers on both sides of the aisle, ready to whisper in ears when votes come up.
That’s not cynicism. That’s just how this works. And now Anthropic is working it too.
The “safety-first” AI company has entered the influence-buying business. They’ll tell you it’s about ensuring smart regulation. About making sure policymakers understand the technology. About protecting innovation.
Sure. And I’ve got a bridge to sell you.
đź•’ Published: