The public standoff between the Trump administration and Anthropics has reached a stalemate Military officials demand The artificial intelligence company is bending its ethics policies by Friday or risk hurting its business.
Anthropic CEO Dario Amodei drew a sharp red line 24 hours before the deadline, declaring that his company “cannot in good conscience respond” to the Pentagon’s eventual demand to allow unrestricted use of its technology.
Anthropic, maker of chatbot Cloud, can afford to lose the defense contract. But Defense Secretary Pete Hegseth’s ultimatum this week posed broader risks at the height of the company’s meteoric rise from little-known computer science research lab in San Francisco to one of the world’s most valuable startups.
If Amodei doesn’t budge, military officials have warned that they will not only pull Anthropic’s contract but also view it as a supply chain risk, a label often stamped on foreign adversaries that could derail the company’s important partnerships with other companies.
If Amodei gives up, he could lose confidence in the burgeoning AI industry, especially from top talent drawn to the company because of its promise of responsibly building better-than-human AI, which could pose catastrophic risks without safeguards.
Anthropic said it had requested limited assurances from the Pentagon that CLOUD would not be used for mass surveillance of Americans or in fully autonomous weapons. But after months of private conversations turned public, she said in a statement Thursday that the new contract language “framed as a compromise was paired with laws that would allow those safeguards to be ignored at will.”
That was after Sean Parnell, the Pentagon’s chief spokesman, posted on social media that “we will not let any company dictate the terms of how operational decisions are made” and added that the company “has until 5:01 p.m. ET on Friday to decide” whether it will meet the demands or face the consequences.
Later, Emil Michael, Under Secretary of Defense for Research and Engineering, criticized Amodei, claiming on
That message did not resonate in much of Silicon Valley, where a growing number of tech workers from Anthropic’s biggest rivals, OpenAI and Google, expressed support for Amodei’s position late Thursday in an open letter.
OpenAI and Google, along with Elon Musk’s xAI, also have contracts to supply the military with their AI models.
“The Pentagon is negotiating with Google and OpenAI to try to convince them to agree to what Anthropic rejected,” the open letter said. “They are trying to divide each company out of fear that the other will give up.”
Republican and Democratic lawmakers and a former leader of the Department of Defense’s artificial intelligence initiatives have also raised concerns about the Pentagon’s approach.
“Making a central point on anthropology gets hot headlines, but everyone loses in the end,” retired Air Force Gen. Jack Shanahan wrote in a social media post.
Shanahan faced a different wave of opposition from tech workers during the first Trump administration when he led Project Maven, a project that used artificial intelligence technology to analyze drone footage and target weapons. Many Google employees protested its participation in Project Maven at the time and after that the tech giant refused to renew the contract Pledge not to use artificial intelligence In weapons.
“Since I was in the middle of a Maven project & “Google, it’s reasonable to assume I’d side with the Pentagon here,” Shanahan wrote Thursday on social media. “However, I’m more sympathetic to Anthropic’s position than I was at Google in 2018.”
He said Claude is already widely used across government, including in secret places, and that the red lines set by Anthropic are “reasonable.” He said the large language models of AI that power chatbots like Cloud “are not ready for prime-time use in national security settings,” especially for fully autonomous weapons.
“They’re not trying to play nice here,” he wrote.
Parnell confirmed on Thursday that the Pentagon wants “ Use the anthropic model For all lawful purposes,” he said that opening up the use of the technology would prevent the company from “jeopardizing critical military operations,” though neither he nor other officials would clarify how they wanted to use the technology.
The military “has no interest in using AI to conduct mass surveillance of Americans (which is illegal), nor do we want to use AI to develop autonomous weapons that operate without human intervention,” Parnell wrote.
When Hegseth and Amodei met on Tuesday, military officials warned that they could be appointed Anthropic As a supply chain risk, canceling their contract or invoking a Cold War-era law It’s called the Defense Production Act To give the military greater authority to use its products, even if the company doesn’t approve.
“These last two threats are contradictory in nature: one classifies us as security risks; the other describes Claude as essential to national security,” Amodei said Thursday. He said he hopes the Pentagon will reconsider in light of Claude’s importance to the military, but if not, Anthropic will “enable a smooth transition to another provider.”
—-
AP correspondent Konstantin Torobin contributed to this report.