President Trump directed every federal agency to cease using Anthropic's artificial intelligence technology after the company's CEO publicly objected to how the Pentagon intended to deploy its models. The move, announced via Truth Social, came one day after Anthropic CEO Dario Amodei outlined what he characterized as red lines in the company's relationship with the Department of War.

Secretary of War Pete Hegseth followed through on Feb. 27, designating Anthropic "a supply chain risk to national security" and severing the firm's Pentagon contracts entirely. Anthropic has since filed a lawsuit claiming the designation violates its First Amendment rights, seeking a court order to reverse its blacklisting and restore its federal contracts.

The escalation was swift. But the fault lines had been visible for a long time.

What Amodei Said, and What It Cost

On Feb. 26, just ahead of the launch of Operation Epic Fury, Amodei went public with what he described as his company's discussions with the Department of War. He flagged two categories of use he said should not be included in Pentagon contracts: "mass domestic surveillance" and "fully autonomous weapons."

Whether those concerns are reasonable in the abstract is beside the point. Amodei chose to draw those lines publicly, days before a military operation, framing them as moral boundaries the Pentagon could not cross. The administration read it as an AI company attempting to dictate the terms under which the United States defends itself. Fox News shares.

Trump's response left no room for ambiguity:

"THE UNITED STATES OF AMERICA WILL NEVER ALLOW A RADICAL LEFT, WOKE COMPANY TO DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS! That decision belongs to YOUR COMMANDER-IN-CHIEF, and the tremendous leaders I appoint to run our Military."

Hegseth drove the point further:

"The Terms of Service of Anthropic's defective altruism will never outweigh the safety, the readiness, or the lives of American troops on the battlefield."

That phrase, "defective altruism," was not an accident. It's a direct shot at the "effective altruism" philosophy that permeates Anthropic's corporate identity. Hegseth made clear the Department of War would not negotiate on the terms of a Silicon Valley ideology.

A Company Stacked With Democratic Operatives

The administration's frustration didn't materialize out of thin air. A look at Anthropic's policy and communications apparatus reveals a hiring pattern that reads less like a nonpartisan tech firm and more like a Democratic alumni network.

Consider the roster:

• Sarah Heck, announced earlier this month as Head of Public Policy, previously worked on the National Security Council during the Obama administration and served as an advisor to the Obama Foundation. In 2020, she posted an article by then-Maryland Gov. Larry Hogan with the caption hashtag "#DumpTrump."
• Brian Peters, head of North America Government Affairs, previously worked for former Rep. Jay Inslee, D-Wash., who would later become governor of the state.
• Maxwell Young, head of policy communications, previously worked for Sen. Chuck Schumer and handled communications for former New York City Mayor Eric Adams.
• Maya Humes, who handles policy communications, served as North Carolina communications director for the 2020 Biden-Harris campaign and went on to work in the Biden White House and State Department for the entirety of his term.
• Chris Nulty, Head of External Communications, spent 10 years at Airbnb as a senior communications director and reposted tweets supporting the deplatforming of Trump after Jan. 6, including a post from Hillary Clinton.

Anthropic does employ some people with Republican credentials. Jared Powell previously served in the offices of Rep. Laurel Lee, R-Fla., and former Rep. Cathy McMorris Rodgers, R-Wash. Mary Croghan served in the first Trump administration. Leah Graham served in the George W. Bush administration and worked at the Republican National Committee. Former Trump administration official Chris Liddell sits on the board of directors, and former Sen. Roy Blunt, R-Mo., serves on Anthropic's National Security and Public Sector Advisory Council.

But the center of gravity is unmistakable. The people shaping the company's public policy posture and communications strategy are overwhelmingly drawn from Democratic politics. When your head of public policy posted "#DumpTrump" and your comms lead reposted calls to deplatform the sitting president, the claim of nonpartisan good faith requires more than a mission statement to sustain.

Follow the Money

The personnel problem extends to the investor class. Netflix co-founder Reed Hastings, who is tied to Anthropic, gave $20 million to Democrats, including $7 million to a pro-Harris super PAC, according to the Washington Examiner. Facebook co-founder Dustin Moskovitz, an Anthropic backer, reportedly gave $38 million to a Harris-aligned super PAC in 2024.

Amodei himself reportedly endorsed then-Vice President Kamala Harris in 2024. According to The Information, a now-deleted 2018 Facebook post from Amodei called Trump a "serious and legitimate threat to the rule of law."

This is the man who then asked the administration to trust his company with Pentagon contracts while his team publicly set conditions on how the military could use the technology.

The Leaked Memo and the Apology Tour

A leaked internal memo, first published by The Information earlier this month, reportedly showed Amodei suggesting that the Trump administration was retaliating because the company failed to give the president "dictator-style praise." Amodei later stated he "did not leak this post nor direct anyone else to do so" and said it was "not in our interest to escalate this situation."

The apology came too late and landed too softly. Accusing an administration of demanding dictatorial flattery, even in an internal memo, is not a neutral observation. It's an indictment dressed as analysis. The fact that it leaked only confirmed for the White House what they already suspected about the company's internal culture.

In October 2025, Amodei issued a public statement affirming Anthropic's commitment to American AI leadership and arguing that managing AI "should be a matter of policy over politics." Earlier, in October 2024, he published an essay titled "Machines of Loving Grace: How AI Could Transform the World for the Better," for which he credited prominent liberal writer Matt Yglesias as among those who reviewed drafts and provided feedback.

Policy over politics is a fine slogan. But when your reviewer list, your donor base, your policy staff, and your CEO's public endorsements all point in the same political direction, the slogan rings hollow.

The National Security Counterargument

Not everyone on the right agrees with the decision to sever ties. Josh Hodges, who served as House Speaker Mike Johnson's national security advisor and sits on Anthropic's advisory council, pushed back in comments to Fox News Digital:

"Their tools have been used in cyber defense and other missions critical to national security, including Venezuela, including Iran, arguably two of the most complex operations in recent history, according to press reports, of course. It's the only company that's walked away from hundreds of millions in revenue, cutting off ties to CCP-linked firms. In my opinion, that's not a woke company. That's a company that is putting America first."

Hodges also offered a pointed warning:

"I agree it's reasonable to call out companies that are showing an active bias, but cutting off America's world-leading AI company over vibes, hands China a win we can't afford right now."

It's a serious argument, and it deserves a serious hearing. The AI race with China is real, and Anthropic is by many accounts among the most capable American firms in the space. Walking away from hundreds of millions in CCP-linked revenue is nothing.

But the counterargument writes itself. A company that publicly conditions its cooperation with the U.S. military, whose leadership has a documented history of opposing the sitting president, and whose internal communications characterize administration expectations as dictatorial, is not a reliable defense partner. Capability without trust is a liability, not an asset.

The Deeper Pattern

Anthropic's dilemma is not unique. It reflects a broader tension in Silicon Valley, where companies built on progressive cultural assumptions suddenly discover they need the goodwill of a government that does not share those assumptions.

The playbook is familiar. Build a company saturated in left-leaning politics. Hire from the Democratic ecosystem. Let your CEO endorse the Democratic presidential candidate. Then, when the other party holds power, insist you're nonpartisan and demand access to government contracts on your own terms.

Heck's "#DumpTrump" post was from 2020. Nulty's reposting of deplatforming calls happened in the wake of Jan. 6. Amodei's deleted Facebook post was from 2018. None of this is ancient history. These are the people now asking the Trump administration to believe that Anthropic operates in good faith.

Anthropic says it wants to ensure AI benefits everyone. It says it wants to maintain America's lead. It even took the step, as Heck stated, of committing to cover "100% of electricity price increases that consumers face from our data centers" in support of the White House ratepayer protection pledge. These are real gestures.

But gestures don't erase a pattern. And patterns matter when the stakes are national security.

What Comes Next

Anthropic's lawsuit will test whether the "supply chain risk to national security" designation can survive judicial scrutiny. The company argues its First Amendment rights were violated. The administration will argue that no company has a constitutional right to a Pentagon contract, especially one whose leadership actively undermined confidence in the partnership.

The courts will sort out the legal question. The political question is already settled. When you spend years signaling that you find the commander-in-chief dangerous, illegitimate, and unworthy of anything short of opposition, you don't get to act surprised when he takes you at your word.