The AI 'Safety Company' Is Now Buying Congress. Anthropic Just Filed to Create Its Own PAC.
Anthropic filed to create AnthroPAC while AI companies pour $300M+ into the 2026 midterms. The safety-first lab now plays the same Washington game as everyone else.
The AI Post newsroom — delivering AI news at the speed of intelligence.
Anthropic, the company that built its entire brand on being the responsible AI lab, just filed paperwork with the Federal Election Commission to create AnthroPAC. An employee-funded political action committee. Bipartisan donations to House and Senate candidates. Maximum $5,000 per employee per year.
In other words: the safety company is buying politicians now. Just like everybody else.
The timing is not subtle. AI companies have collectively poured more than $300 million into the 2026 midterm races. That number, reported by the Washington Post, makes Big Tech's previous political spending look like pocket change. And Anthropic is not just running a small employee PAC. In February, the company donated $20 million to Public First Action, a super PAC backing candidates who support AI regulation. That pushed Public First's fundraising target from $50 million to $75 million.
Let that sink in. The company that warns governments about existential AI risk is simultaneously spending tens of millions to make sure those governments write the rules Anthropic wants.
This is not necessarily hypocrisy. It might be survival. Anthropic is locked in a vicious legal battle with the Pentagon, which designated it a supply chain risk earlier this year. The company needs allies in Congress who understand AI policy and will push back on heavy-handed government action. A PAC is how you get those allies.
Google, Microsoft, Amazon, and Meta already run identical employee-funded PACs. Anthropic is just copying the playbook. The difference is that none of those companies built their reputation on being above the game. Anthropic did. And now it is very much in the game.
Here is the real question nobody is asking: if $300 million in AI industry money flows into one midterm cycle, who exactly is representing the public interest? The candidates taking the checks? The regulators those candidates appoint? The AI safety researchers who work at the companies funding the PACs?
The FEC filing lists Allison Rossi as treasurer, signed from Anthropic's San Francisco headquarters. Jared Powell serves as assistant treasurer. All perfectly legal. All perfectly normal for a tech company in 2026. And that is exactly the problem. Anthropic was not supposed to be normal.
First reported by The Hill and TechCrunch, with additional reporting from Bloomberg.