Anthropic Opens Its Wallet in Washington: Inside the AI Maker's New Political Action Committee

Anthropic, the San Francisco-based artificial intelligence company behind the Claude chatbot, has formally registered a political action committee -- a move that signals a dramatic escalation in the company's willingness to play the influence game in Washington. The filing, first reported by TechCrunch, marks a turning point for a firm that has long positioned itself as the safety-conscious alternative to rivals like OpenAI and Google DeepMind.

The PAC, registered with the Federal Election Commission, will allow Anthropic to pool employee contributions and direct funds to political candidates sympathetic to its policy priorities. It's a well-worn tactic in corporate America -- but a relatively fresh one for AI startups, most of which have historically preferred to exert influence through lobbying shops, think-tank donations, and quiet conversations with staffers on Capitol Hill.

Not anymore.

Anthropic's decision to stand up a PAC comes at a moment when the regulatory environment for AI in the United States is shifting fast. Congress has spent the better part of two years debating how -- and whether -- to impose binding rules on foundation model developers. Multiple bills are circulating in both chambers, ranging from narrow disclosure requirements to sweeping licensing regimes that would hand federal agencies significant oversight authority over the largest AI systems. For a company valued at roughly $60 billion following its latest funding round, the stakes couldn't be higher.

The company has been quietly building out its government affairs operation for months. It hired its first dedicated lobbyists in 2024 and opened a Washington office shortly afterward. Federal lobbying disclosure records show that Anthropic's spending on lobbying more than tripled between 2024 and 2025, though the totals remain modest compared with tech giants like Google parent Alphabet or Meta, each of which spends tens of millions annually. The PAC represents the next logical step -- a way to put money directly behind the candidates Anthropic believes will shape AI policy most favorably.

But the move also carries reputational risk. Anthropic has cultivated a public identity rooted in caution and responsibility. Its founders, Dario and Daniela Amodei, left OpenAI in 2021 partly over disagreements about safety practices, and the company has published extensively on AI alignment research, interpretability, and what it calls "responsible scaling." Launching a PAC -- an instrument of raw political power -- sits uneasily alongside that image. Critics have already begun pointing out the tension.

"There's something dissonant about a company that says it's building the most dangerous technology in human history also spending money to influence the politicians who are supposed to regulate it," said one AI policy researcher at a Washington think tank, who asked not to be named because they work with multiple AI companies. The concern isn't unique to Anthropic. It applies to every AI firm now wading into electoral politics. But Anthropic's brand makes the optics sharper.

The company, for its part, has framed the PAC as a natural extension of its policy engagement. In a statement reported by TechCrunch, Anthropic said it wants to support candidates "who understand the importance of AI safety and American competitiveness in artificial intelligence." That dual emphasis -- safety and competitiveness -- reflects a messaging strategy the company has refined over the past year. It allows Anthropic to appeal simultaneously to Democrats worried about AI harms and Republicans focused on keeping the U.S. ahead of China in the technology race.

It's a shrewd formulation. And it mirrors the broader lobbying posture of the AI industry, which has increasingly wrapped its policy preferences in national security language. The argument goes like this: overly burdensome regulation will slow down American AI companies while Chinese competitors, unburdened by democratic constraints, race ahead. Therefore, Congress should regulate lightly and invest heavily. Anthropic hasn't stated it quite so bluntly, but the subtext of its PAC's mission statement is hard to miss.

Anthropic is not the first AI-focused company to enter the PAC arena. OpenAI has also ramped up its political spending, and several industry trade groups -- including the Information Technology Industry Foundation and the newer AI-focused lobbying coalitions -- have been active in campaign finance for years. What makes Anthropic's entry notable is the speed. The company is barely five years old. Most startups at this stage are still figuring out their go-to-market strategy, not registering political committees.

Then again, most startups aren't sitting at the center of a global debate about existential risk.

The timing also coincides with a broader wave of AI-related political activity. The 2026 midterm elections are shaping up to be the first cycle in which AI policy features prominently in campaign messaging. Several Senate candidates have made AI regulation a plank of their platforms, and at least two House races have seen significant spending by tech-aligned super PACs. Anthropic's PAC gives the company a seat at that table -- a way to reward allies and, implicitly, signal consequences to opponents.

How much money the PAC will raise remains to be seen. Corporate PACs typically draw contributions from employees, often senior executives, and the amounts tend to be modest in the context of federal elections. A PAC raising a few hundred thousand dollars per cycle won't rival the war chests of major industry groups. But the symbolic value is significant. It tells lawmakers that Anthropic is serious about sustained engagement -- that it isn't going away after one hearing or one bill.

The company's lobbying priorities offer clues about where the PAC's money might flow. Anthropic has been particularly active on issues related to AI model evaluation, export controls on AI chips, and federal procurement of AI systems. It has advocated for a regulatory framework that distinguishes between different levels of AI capability -- a tiered approach that would impose the strictest requirements only on the most powerful models. This framework, not coincidentally, would likely benefit Anthropic, which competes directly with OpenAI and Google at the frontier of model capability but argues it does so more carefully.

On export controls, Anthropic has generally supported restrictions on selling advanced AI chips to China, a position that aligns it with the Biden-era Commerce Department rules and, more recently, with bipartisan sentiment in Congress. The company has been less vocal about the secondary effects of those controls -- the impact on allied nations, the potential for driving chip manufacturing to less regulated jurisdictions -- but its public statements have consistently emphasized the national security rationale.

Federal procurement is another area of intense focus. The U.S. government is one of the largest potential customers for AI systems, and companies that shape procurement standards early stand to gain enormously. Anthropic has pitched its models for use in government contexts, emphasizing their safety features and the company's willingness to submit to third-party audits. A PAC that supports candidates friendly to AI adoption in government agencies could accelerate that effort considerably.

So where does this leave the broader AI policy debate?

The proliferation of AI-linked PACs and lobbying operations has prompted growing concern among civil society groups that the industry is capturing the regulatory process before it even fully begins. Organizations like the Electronic Frontier Foundation and the AI Now Institute have warned that the concentration of lobbying power among a handful of well-funded AI companies risks producing rules that serve corporate interests rather than public safety. The entry of Anthropic -- a company that has genuine credibility on safety issues -- into the PAC world complicates that narrative somewhat. But it doesn't resolve it.

There's also the question of internal dynamics. PACs require employee buy-in, and Anthropic's workforce includes many researchers who came to the company specifically because of its safety mission. Some may welcome the PAC as a way to amplify that mission politically. Others may view it as a corruption of the company's founding principles. The tension between Anthropic's research culture and its growing corporate ambitions has been a recurring theme in industry circles, and the PAC is likely to intensify it.

Dario Amodei has addressed this tension obliquely in past public remarks. In a widely circulated essay last year, he argued that AI companies have a responsibility to engage with government -- not just through research publications, but through active policy advocacy. "If the people building these systems don't participate in shaping the rules, the rules will be shaped by people who don't understand the technology," he wrote. It's a reasonable argument on its face. But it's also the same argument every regulated industry has made since the dawn of modern lobbying.

The PAC's formation also arrives against the backdrop of Anthropic's rapidly growing commercial ambitions. The company recently expanded its enterprise offerings, struck cloud partnerships with Amazon Web Services and Google Cloud, and launched new versions of its Claude model aimed at business users. Revenue has grown sharply -- reportedly approaching an annualized run rate of several billion dollars -- and the company is competing aggressively for market share in sectors like finance, healthcare, and legal services. Political engagement is, in this context, simply another front in a multi-front competitive war.

And competition is fierce. OpenAI, backed by Microsoft, has its own extensive government relations operation and has been courting defense and intelligence agencies. Google DeepMind benefits from Alphabet's massive lobbying infrastructure. Meta has taken a different tack, open-sourcing its Llama models and arguing that open-source AI should face lighter regulation -- a position that conveniently disadvantages its closed-source competitors. In this environment, Anthropic can't afford to be the only major player without a PAC.

That doesn't mean the decision was inevitable. Plenty of companies resist the pull of political spending, at least for a time. But Anthropic's leadership appears to have concluded that the window for shaping AI regulation is narrowing, and that passive engagement -- white papers, testimony, op-eds -- isn't enough. A PAC is a blunter instrument. It's also a more effective one.

The next few months will reveal the PAC's initial fundraising totals and its first disbursements. Those details will matter. Which candidates receive money, in which races, and at what stage of the election cycle will tell us far more about Anthropic's political strategy than any press release. If the PAC funds candidates across party lines who share a common interest in AI competitiveness and light-touch regulation, it will confirm what many observers already suspect: that Anthropic's political identity is fundamentally pragmatic, not ideological.

For Washington, the message is clear. The AI industry isn't just coming to town. It's moving in, hiring staff, opening offices, and now writing checks. Anthropic's PAC is one more data point in a trend that has been accelerating for two years. The companies building the most powerful AI systems in the world have decided that political power is not optional. It's infrastructure.

Whether that's good for democracy depends on who you ask. What's not in dispute is that it's happening -- and that Anthropic, for all its talk of safety and caution, has decided it would rather be inside the room than outside it.