(Photo by Jaque Silva/NurPhoto via Getty Images)
Anthropic took a stand against the Pentagon in February, giving up a $200 million contract rather than loosening the safety guardrails on its artificial intelligence model, Claude. The decision was on brand for a company that has gone on the offensive with a surge in lobbying and political spending in support of AI regulation.
In 2025, Anthropic PBC quadrupled its federal lobbying spending to more than $3.1 million and, last month, directed $20 million to a political nonprofit whose affiliated political action committees poured millions into House primaries just two weeks before voting began in a handful of states.
The Trump administration has butted heads with Anthropic’s leaders previously, seeing their advocacy for regulation and warnings about AI’s impact on jobs as a nuisance to the White House’s policy agenda. Trump’s AI czar, David Sacks, has accused Anthropic of “running a sophisticated regulatory capture strategy based on fear-mongering” and labeled the company as “woke.”
On Feb. 27, its $200 million contract with the Pentagon fell apart after tense negotiations over whether the company would loosen its AI safety policies and guardrails to allow the military unfettered use of its models. Pentagon officials, who were seeking to be able to use the model for “all lawful purposes,” took issue with Anthropic’s demands that their technology not be used in the mass surveillance of American citizens or to empower autonomous weapons.
Anthropic is a public benefit corporation and has AI safety as a guiding principle, building safeguards into its models and frequently advocating for stricter regulation. The company has positioned AI safety as a core part of its identity and has spent millions of dollars to advocate for tougher regulation of AI companies and technology.
Emerging technologies researcher Jeffrey Ding said Anthropic’s noncompliance with the Pentagon reflects the company's concern over AI safety.
“This action has revealed that they are willing to put a lucrative contract on the line to signal and to substantiate their commitment to building safer AI models,” Ding said.
A pair of Super PACs
On Feb. 12, the company announced it was donating $20 million to Public First Action, a new 501(c)(4) calling for tougher regulation of AI companies and technology. Only a week later, one of the group’s super PACs filed its first independent expenditure. After only two weeks, its pair of political committees – Jobs and Democracy, and Defending Our Values – reported $2.8 million in independent expenditures. They have supported candidates running for the House of Representatives in New York, North Carolina and Texas.
Founded by two former House members, Brad Carson (D-Okla.) and Chris Stewart (R-Utah), Public First Action has supported both Democratic (through Jobs and Democracy) and Republican (Defending Our Values) candidates.
So far, Anthropic is the only disclosed donor to Public First Action, but the group said it had raised nearly $50 million in February.
The group is already going up against super PACs backed by other tech executives who favor lighter regulation. Leading the Future PAC, backed by OpenAI’s Greg Brockman and other tech leaders, announced it already raised $125 million to influence the midterm elections.
The Anthropic-backed super PACs began donating to candidates in Texas and North Carolina only two weeks before their March 3 primaries, shifting the spending dynamics in those races at the last minute.
The Democratic arm of the group has spent $1.6 million supporting Rep. Valerie Foushee (D) in the competitive primary for North Carolina’s 4th district. Foushee serves on a newly created House Democratic Commission on AI.
Foushee is running against Nida Allam, who’s getting support from American Priorities, a super PAC created to counter pro-Israel spending. American Priorities, which has spent about $1 million supporting Allam, was the race’s biggest spender before Jobs and Democracy entered the primary race in the last few weeks.
Allam released a new ad on Monday attacking Foushee for accepting Jobs and Democracy’s contribution, mentioning that its backers’ technology was used in the recent strike against Iran.
“As election day approaches, you will see nearly $2 million of ads for my opponent, funded by AI-backed super PACs, the same AI corporation that powered Trump's attacks on Iran,” Allam said in the ad, only a day before the primary.
The Anthropic-backed super PACs spent $864,000 leading up to Texas’ primary, funding three candidates: Carlos De La Cruz (R), Alex Mealer (R) and Colin Allred (D).
Public First and its affiliated PACs have supported two of the three co-chairs of the House Democratic Commission on AI. In addition to Foushee, Rep. Josh Gottheimer (D-N.Y.) received support from Public First, which reportedly spent $300,000 on ads supporting his race.
Alex Bores, who’s running for Congress in New York’s 12th district and has made tech regulation a pillar of his campaign, was the target of attack ads from the OpenAI-backed Leading the Future. In response, Jobs and Democracy spent $295,000 on ads supporting Bores.
Lobbying in favor of regulation
In 2025, Anthropic spent more than $3.1 million lobbying the federal government on issues related to AI, including regulations, national security, export controls, federal procurement and infrastructure. Anthropic quadrupled its spending last year, one of the highest rates of growth among companies in the AI ecosystem.
One of the bills Anthropic lobbied included Trump’s omnibus tax bill, which included a 10-year ban on states regulating AI. Many tech leaders were proponents of the moratorium. OpenAI CEO Sam Altman and Microsoft president Brad Smith both testified to Congress in support of the measure in May.
In contrast, Anthropic CEO Dario Amodei published an op-ed in The New York Times urging against the moratorium and emphasizing the need for AI companies to provide transparency around the safety of their products.
“At the federal level, instead of a moratorium, the White House and Congress should work together on a transparency standard for A.I. companies, so that emerging risks are made clear to the American people,” Amodei said.
Amodei and his company have also advocated extensively for export controls on advanced chips over concerns about Chinese competitors.
“Anthropic is one of the most aggressive, I would say, ‘America first’ type companies in the AI space with their support of export controls,” Ding said. “But now they’ve been singled out from the current administration.”
To help advance its policy division, Anthropic hired several Biden administration officials, such as Tarun Chhabra, who heads the company’s national security policy, and Elizabeth Kelly, who heads its “AI for social good” team.
And it has also leaned on resources with closer ties to the current White House, contracting with Continental Strategy in the second quarter of last year. The company paid $290,000 total in 2025 to Continental Strategy, which former Trump adviser Carlos Trujillo heads.
One of the lobbyists deployed by Continental Strategy was Alberto Martinez, who was Secretary of State Marco Rubio’s chief of staff when Rubio was in the Senate. Martinez lobbied the House, Senate and Office of the President on AI development and national security for Anthropic.
‘Pocket change’ for the inauguration
Tech companies shelled out record donations to Trump’s inaugural committee in 2024, and while Anthropic was no exception, its donation was meager in comparison to its peers.
Anthropic contributed $50,000 to Trump’s inauguration, only a twentieth of what OpenAI, Perplexity AI and C3.ai each gave. Compared to what Anthropic has spent on lobbying and election influence, its donation to Trump’s inaugural committee is pocket change.
By Hien An Ngo