Trump orders federal agencies to stop using Anthropic’s AI after clash with Pentagon

President Trump on Friday directed federal agencies to stop using technology from San Francisco artificial intelligence company Anthropic, escalating a high-profile clash between the AI startup and the Pentagon over safety.

In a Friday post on the social media site Truth Social, Trump described the company as “radical left” and “woke.”

“We don’t need it, we don’t want it, and will not do business with them again!” Trump said.

The president’s harsh words mark a major escalation in the ongoing battle between some in the Trump administration and several technology companies over the use of artificial intelligence in defense tech.

Anthropic has been sparring with the Pentagon, which had threatened to end its $200-million contract with the company on Friday if it didn’t loosen restrictions on its AI model so it could be used for more military purposes. Anthropic had been asking for more guarantees that its tech wouldn’t be used for surveillance of Americans or autonomous weapons.

The tussle could hobble Anthropic’s business with the government. The Trump administration said the company was added to a sweeping national security blacklist, ordering federal agencies to immediately discontinue use of its products and barring any government contractors from maintaining ties with it.

Defense Secretary Pete Hegseth, who met with Anthropic’s Chief Executive Dario Amodei this week, criticized the tech company after Trump’s Truth Social post.

“Anthropic delivered a master class in arrogance and betrayal as well as a textbook case of how not to do business with the United States Government or the Pentagon,” he wrote Friday on social media site X.

Anthropic didn’t immediately respond to a request for comment.

Anthropic announced a two-year agreement with the Department of Defense in July to “prototype frontier AI capabilities that advance U.S. national security.”

The company has an AI chatbot called Claude, but it also built a custom AI system for U.S. national security customers.

On Thursday, Amodei signaled the company wouldn’t cave to the Department of Defense’s demands to loosen safety restrictions on its AI models.

The government has emphasized in negotiations that it wants to use Anthropic’s technology only for legal purposes, and the safeguards Anthropic wants are already covered by the law.

Still, Amodei was worried about Washington’s commitment.

“We have never raised objections to particular military operations nor attempted to limit use of our technology in an ad hoc manner,” he said in a blog post. “However, in a narrow set of cases, we believe AI can undermine, rather than defend, democratic values.”

Tech workers have backed Anthropic’s stance.

Unions and worker groups representing 700,000 employees at Amazon, Google and Microsoft said this week in a joint statement that they’re urging their employers to reject these demands as well if they have additional contracts with the Pentagon.

“Our employers are already complicit in providing their technologies to power mass atrocities and war crimes; capitulating to the Pentagon’s intimidation will only further implicate our labor in violence and repression,” the statement said.

Anthropic’s standoff with the U.S. government could benefit its competitors, such as Elon Musk’s xAI or OpenAI.

Sam Altman, chief executive of OpenAI, the company behind ChatGPT and one of Anthropic’s biggest competitors, told CNBC in an interview that he trusts Anthropic.

“I think they really do care about safety, and I’ve been happy that they’ve been supporting our war fighters,” he said. “I’m not sure where this is going to go.”

Anthropic has distinguished itself from its rivals by touting its concern about AI safety.

The company, valued at roughly $380 billion, is legally required to balance making money with advancing the company’s public benefit of “responsible development and maintenance of advanced AI for the long-term benefit of humanity.”

Developers, businesses, government agencies and other organizations use Anthropic’s tools. Its chatbot can generate code, write text and perform other tasks. Anthropic also offers an AI assistant for consumers and makes money from paid subscriptions as well as contracts. Unlike OpenAI, which is testing ads in ChatGPT, Anthropic has pledged not to show ads in its chatbot Claude.

The company has roughly 2,000 employees and has revenue equivalent to about $14 billion a year.