Anthropic, the company that built its entire brand on being the responsible adults in the AI room, just filed paperwork with the Federal Election Commission to launch AnthroPAC. That is a political action committee. For the midterm elections. Funded by employee contributions capped at $5,000 per person.

Let that sink in for a second. The company that lectures every other lab about AI safety, that fought the Pentagon over ethical use of its models, that positioned itself as the conscience of the industry, is now cutting checks to politicians. Both parties. Current lawmakers and rising candidates. The filing, first reported by Bloomberg, includes a signature from Anthropic treasurer Allison Rossi.

This is not surprising. This is inevitable. And it is still deeply telling.

The AI industry has already poured over $300 million into the 2026 midterms, according to tracking by multiple outlets. The Washington Post reported last month that AI companies had contributed $185 million to midterm races. The New York Times previously revealed that Anthropic had quietly funneled at least $20 million into Public First, a Super PAC financing ad campaigns pushing Anthropic-friendly regulation. AnthroPAC is just the above-board version of what was already happening behind the scenes.

The timing is not subtle. Anthropic is still locked in a brutal legal fight with the Pentagon over how the military uses its AI models. The Defense Department labeled Anthropic a supply chain risk earlier this year. When your biggest potential customer is also your legal adversary, you do not leave Congress to sort it out on their own. You buy a seat at the table.

Here is the real story: every AI company that claimed to be different from Big Tech is now doing exactly what Big Tech has done for decades. Google, Meta, Microsoft, Amazon all have massive lobbying operations and PACs. They learned a long time ago that the most efficient way to shape regulation is to fund the regulators. Anthropic just got the memo.

The employee-funded framing is doing heavy lifting here. "Voluntary contributions capped at $5,000" sounds grassroots. It is not. When your average engineer makes $300,000 a year and your company is about to IPO, $5,000 is a rounding error. This PAC will be extremely well-funded.

The question nobody is asking: what does Anthropic want from Congress? The obvious answer is favorable AI regulation that keeps the Pentagon from weaponizing their models without consent, protects their competitive position against OpenAI, and prevents states from writing patchwork rules that complicate their business. That is a lot to ask. And it costs a lot more than $5,000 per employee.

Watch the other labs. If Anthropic is going bipartisan, OpenAI and Google will match or exceed within weeks. The AI industry midterm spending war just got its newest combatant. And the "safety-first" company just proved that in Washington, safety and self-interest are the same budget line.