Trump claims he dismissed Anthropic “like dogs” as the Pentagon officially restricts the AI startup.

Donald Trump made headlines on Thursday by claiming to have cut off relations between the US military and Anthropic, coinciding with reports that negotiations between the Department of Defense and the AI startup were back on track.

These recent events mark the latest chapter in the complex tug-of-war between the US government and Anthropic.

“Well, I fired Anthropic. Anthropic is in trouble because I fired [them] like dogs, because they shouldn’t have done that,” Trump declared in an interview with Politico on Thursday.

Shortly after Trump’s announcement, the Pentagon officially classified Anthropic as a “supply chain risk,” a designation that prohibits all government contractors from utilizing the startup’s technology. This classification has never been applied before to a US-based company.

“DoW officially informed Anthropic leadership that the company and its products are deemed a supply chain risk, effective immediately,” a Pentagon official stated in comments made to Bloomberg.

Contrary to this alarming news, reports also indicate a potential easing of tensions. Sources from the Financial Times and Bloomberg revealed that discussions between the Pentagon and Anthropic regarding the military’s use of its AI technologies and contracts had recommenced. Anthropic’s offerings, which include the widely used Claude chatbot and coding assistant, have already been integrated into Palantir’s Maven system—an essential asset for military intelligence that played a role in recent operations against Iran, according to the Washington Post.

Dario Amodei, CEO of Anthropic, has been actively negotiating the Pentagon’s contract with Emil Michael, the undersecretary of defense for research and engineering, as noted by Bloomberg. However, the dynamics between the two are fraught, with reports indicating mutual disdain between them, according to the New York Times.

Just last week, Trump ordered a complete halt to the use of Anthropic’s technology by the entire federal government after the company rejected a deal over concerns regarding potential uses of its models for domestic surveillance or autonomous weaponry.

Both the State and Treasury Departments had already begun to sever connections with Anthropic, their heads confirmed. In a related move, Pete Hegseth, the US defense secretary, had verbally labeled Anthropic as a “supply chain risk” last week, prompting Amodei to announce plans for legal action against such a designation.

In a recent internal communication to his staff, Amodei issued a stinging critique of the Trump administration and competitor OpenAI, which quickly stepped in to fill the gap left by the discord between the government and Anthropic, unveiling its own military contract. In this controversial discourse, Amodei described Sam Altman, OpenAI’s CEO, as “mendacious,” dismissing Altman’s assurances to employees that the Pentagon would enforce specific safeguards as mere “safety theater.” Altman, for his part, acknowledged this week in a private message that OpenAI would have no way to supervise how the military might employ its technology.

Amodei commented on the leaked internal message in an official statement on Thursday.

“I want to extend my apologies for a post that was internal to the company and was leaked to the press yesterday. Anthropic did not leak this post nor direct anyone else to do so; it is not in our interest to escalate this situation,” Amodei clarified. “It was a challenging day for us, and I regret the tone of my message. It does not accurately reflect my thought-out or considered views. This communication was created six days ago and is an outdated perspective on our current circumstances.”

The punitive designation against Anthropic could spell considerable financial difficulties for the company. Recent reports put its valuation at around $60bn, which is now threatened due to ongoing tensions with the Pentagon, as detailed by Axios. The larger tech community in Silicon Valley, including OpenAI, has rallied in support of Anthropic during this challenging time.

A spokesperson from Microsoft noted that company attorneys have evaluated the “supply chain risk” designation and concluded that Anthropic’s products, including Claude, can still be available to clients (excluding the Department of War) through platforms like M365, GitHub, and Microsoft’s AI Foundry.

Furthermore, Microsoft will continue its collaboration with Anthropic on projects unrelated to defense, the spokesperson confirmed.

As the Secretary of Defense announced measures against Anthropic, a new deal with OpenAI was also struck to utilize its technology for military operations within classified networks. This timing raised ethical questions and ignited public backlash, resulting in Claude’s app reaching the top of download charts across the US. In the days that followed, Altman expressed his willingness to revise the agreement with the Defense Department, conceding that his company’s actions were perceived as “opportunistic and sloppy.”