Trump blacklists “leftwing nutjob” AI company that drew line at weapons and surveillance

Trump orders agencies to drop Anthropic

President Trump directed every federal agency to stop using Anthropic’s AI technology on Feb. 27, 2026.

He posted the order on Truth Social, calling the company’s leaders “leftwing nut jobs” who tried to strong-arm the military.

Trump said Anthropic’s stance put troops in danger and national security at risk. Agencies have six months to phase out all Anthropic products.

Trump warned the company to cooperate or face “the full power of the presidency,” including civil and criminal consequences.

Two red lines started the fight

The dispute came down to two safeguards Anthropic held firm on in its Pentagon contract for its AI model Claude: no mass domestic surveillance of Americans and no fully autonomous weapons.

The Pentagon wanted the right to use Claude for all lawful purposes without company-imposed limits.

Anthropic said the Pentagon’s final contract language looked like a compromise but included legal wording that would let the safeguards be ignored at will.

The company said neither restriction had ever blocked a single government mission.

Anthropic already ran on classified networks

Anthropic was the first AI company to put a model on the Pentagon’s classified networks. The contract was worth up to about $200 million.

Claude handled intelligence assessments, military planning, and other sensitive work across national security agencies.

Military officials had called Anthropic’s AI the most advanced and secure model for sensitive applications.

That makes the breakup more than symbolic, since the technology already sits deep inside the government’s most protected systems.

Hegseth calls Anthropic a national security risk

Defense Secretary Pete Hegseth took a step no one had taken before with an American company.

He designated Anthropic a supply chain risk to national security, a label historically saved for foreign adversaries.

The designation bars every contractor, supplier, and partner doing business with the U.S. military from any commercial activity with Anthropic.

Hegseth said the military’s warfighters would never be held hostage by what he called the ideological whims of Big Tech.

Anthropic CEO refuses to back down

CEO Dario Amodei said on Feb. 26 that Anthropic could not “in good conscience” accept the Pentagon’s demands.

He said the company understands the military, not private companies, make military decisions, but that some AI uses can undermine democratic values.

Amodei argued that current AI models are not reliable enough for fully autonomous weapons and that using them this way would put warfighters and civilians at risk.

Anthropic vowed to challenge the supply chain risk designation in court, calling it legally unsound and a dangerous precedent.

Pentagon officials fire back at Amodei

Pentagon Undersecretary for Research and Engineering Emil Michael accused Amodei of lying and having a “God-complex” in a post on X.

Michael said Amodei wanted to personally control the U.S. military and was putting national safety at risk.

Pentagon spokesman Sean Parnell wrote on social media that the military would not let any company dictate its operational decisions.

The Pentagon said it had no plans to use AI for mass surveillance or autonomous weapons but would not accept company-imposed limits on lawful military use.

OpenAI lands Pentagon deal hours later

Hours after Anthropic’s ban, OpenAI CEO Sam Altman announced a deal to put OpenAI’s models on the Pentagon’s classified network.

Altman said the deal included bans on domestic mass surveillance and required human responsibility for all use-of-force decisions, including for autonomous weapons.

OpenAI added a third safeguard banning use for high-stakes automated decisions like social credit systems. Altman admitted the timing looked bad and called the deal “definitely rushed.”

Multiple observers noted the Pentagon appeared to accept from OpenAI the same safeguards it punished Anthropic for requesting.

Tech workers rally behind Anthropic

More than 430 employees at Google and OpenAI signed an open letter titled “We Will Not Be Divided” supporting Anthropic’s position.

The letter urged company leaders to stand together against the Pentagon’s demands and alleged that the Pentagon was negotiating with Google and OpenAI to get them to agree to what Anthropic refused.

Google DeepMind chief scientist Jeff Dean publicly opposed mass surveillance, saying it violates the Fourth Amendment.

More than 100 Google DeepMind employees also sent an internal letter urging the company not to allow military surveillance or the use of autonomous lethal weapons.

Bipartisan senators push both sides to talk

Senate Armed Services Committee Chair Roger Wicker and Ranking Member Jack Reed, along with Defense Appropriations Chair Mitch McConnell and Ranking Member Chris Coons, sent a private letter urging both sides to work things out, according to reports.

The letter called for the Pentagon to extend its deadline.

It did not. Republican Sen. Thom Tillis called the public nature of the dispute “sophomoric” and said Congress needs to weigh in on mass surveillance.

Democratic Sen. Mark Warner said the directive raises serious concerns about whether political considerations are driving national security decisions.

Congress starts drafting new legislation

Democrats in both chambers began preparing bills in response.

Rep. Sam Liccardo of California plans to introduce an amendment to the Defense Production Act that would ban federal agencies from retaliating against AI and tech vendors, according to reports.

Sen. Ron Wyden vowed to fight the ban and said he would seek bipartisan support. Sens.

Elizabeth Warren and Andy Kim argued the Defense Production Act was never meant to force companies to help the military conduct surveillance or build autonomous weapons.

The United States currently has no comprehensive federal laws governing AI systems or consumer privacy.

Anthropic plans legal fight as Claude stays online

Anthropic said it would challenge the supply chain risk designation in court.

The company argued Hegseth does not have the legal authority to ban military contractors from using Anthropic products for non-military work.

Hours after the ban, the U.S. military carried out strikes on Iran that relied on Claude for intelligence assessments and battle simulation, according to the Wall Street Journal.

Claude remains available to more than 100,000 users on classified government networks during the six-month phase-out.

Experts warn that replacing Anthropic’s deeply embedded AI across classified networks will be far more complex and costly than six months allows.

The bigger questions for all Americans

This fight raises a fundamental question: can the government force a private company to remove safety restrictions from powerful AI technology?

The mass surveillance safeguard Anthropic defended is rooted in Fourth Amendment protections against unreasonable searches.

Labelling an American company a supply chain risk for refusing government demands could set a precedent affecting any company that negotiates with the federal government.

The dispute could shape future laws on how AI is used by the military and law enforcement. This story is still developing.

This article was created with AI assistance and human editing.

Read more from this brand:

• Pentagon just proved it can fly a nuclear reactor anywhere in the country
• Young Americans Are Enlisting Again, Military Signs 160,000+ Recruits
• Instead of retiring sooner, workers may need to reskill repeatedly in the AI era, analysts say