Anthropic says it’s in ‘productive’ discussions with Pentagon about Claude but still plans to sue on ban – News

Anthropic PBC has shared an update about its artificial intelligence safety feud with the Pentagon.

In a Thursday blog post, Chief Executive Dario Amodei stated that the company has been “having productive conversations” with defense officials. He also addressed several other aspects of the disagreement. 

Last June, Anthropic PBC won a contract to provide the Pentagon with access to its AI models. The company stated that it wouldn’t allow its software to be used for domestic surveillance or the development of fully autonomous weapons. In response, U.S. President Donald Trump ordered federal agencies to stop using Claude.

The first focus of Amodei’s Thursday blog post was not Trump’s order, but rather a related move by U.S. Defense Secretary Pete Hegseth. Hegseth instructed the Pentagon to designate Anthropic as a supply chain risk. Such a designation limits the manner in which U.S. military suppliers can use the affected company’s products.

Amodei wrote in the Thursday blog post that the move won’t affect the “vast majority” of Claude users. Furthermore, it’s expected to have a limited impact on the customers who do have to make adjustments in response to the directive.

“The law requires the Secretary of War to use the least restrictive means necessary to accomplish the goal of protecting the supply chain,” Amodei wrote. “Even for Department of War contractors, the supply chain risk designation doesn’t (and can’t) limit uses of Claude or business relationships with Anthropic if those are unrelated to their specific Department of War contracts.”

The affected suppliers include Microsoft Corp. and Google LLC. The companies have integrated Claude with several of their products, some of which are used by federal agencies. The companies stated this week that they will continue offering Claude access to customers other than the Pentagon.

Anthropic plans to challenge its designation as a supply chain risk in court. The company stated last Friday that implementing the designation “would both be legally unsound and set a dangerous precedent for any American company that negotiates with the government.”

In addition to Anthropic’s regulatory status, Amodei’s blog post addressed a leaked internal memo he sent on Tuesday. In the note, the executive attributed the company’s feud with the Pentagon to his refusal to give “dictator-style praise to Trump.” He also criticized rival OpenAI, which announced a contract with the Pentagon shortly after Anthropic was designated a supply chain risk. 

“It was a difficult day for the company, and I apologize for the tone of the post,” Amodei wrote.

The executive added that Anthropic has been having “productive conversations” with defense officials in recent days. Amodei wrote that the talks are focused on potential ways the company could continue working with the Pentagon in compliance with its AI safety policy. Additionally, Anthropic is working with officials to “ensure a smooth transition” in the event that Claude is phased out of public sector networks. The Trump administration has given federal agencies six months to switch to a different AI provider.

Image: Anthropic