In contract negotiations between senior Defense Department officials and leaders of AI giant Anthropic in December, the company agreed to allow the US government to use its AI systems for missile and cyber defense purposes, said a person familiar with the matter, who requested anonymity to discuss private discussions.
But that apparently didn’t satisfy the Pentagon.
Following weeks of tension between the Department of Defense and Anthropic over the company’s restrictions on how its products can be used by the military, Defense Secretary Pete Hegseth issued a stern ultimatum to the company’s CEO Dario Amodei on Tuesday: Allow AI technology to be used for all legitimate military purposes Friday or be forced to cooperate, a senior Pentagon official told NBC News.
The deadline, explained to NBC News by a senior Pentagon official, comes as Anthropic – a company that heavily markets its focus on AI security – tries to maintain strict policies that prevent its systems from being used for large-scale home surveillance or the direct use of lethal autonomous weapons.
Changes to the December contract will allow its systems to be used more widely for cyber and missile defense, according to a person familiar with the matter. An Anthropic spokesperson told NBC News in a statement that “All iterations of the proposed contract language will enable our models to support missile defense and similar applications.”
But the company’s insistence on roadblocks continued to be a source of contention between Anthropic and the Department of Defense.
According to a senior Pentagon official, representatives of the department, including Undersecretary of Defense Emil Michael, recently discussed several scenarios with Anthropic leadership regarding how the company’s products could be used by the military.
As part of those discussions, officials discussed how Anthropic’s systems could be used if an enemy launched an intercontinental ballistic missile in the U.S. According to a Pentagon source, officials discussed whether Anthropic’s guardrails could somehow prevent a U.S. response to a launch. Anthropic officials said they could be asked to remove those restrictions, according to the official, but Pentagon leadership was not fully satisfied with Anthropic’s changes and did not want to be seen as a private company.
According to an Anthropic spokesperson, any suggestion that CEO Amodei said the Pentagon would have to call on the company for individual missile defense work is “blatantly false.”
In recent discussions, at a meeting on Tuesday, Pentagon leaders said they would invoke the Defense Production Act to force Anthropic to comply with Pentagon regulations, according to a senior Pentagon official. The law allows the president to take control of domestic companies vital to national security in times of need.
At Tuesday’s meeting, Pentagon leadership repeated threats to instead label Anthropic a “supply chain threat” and ban all defense business with the company if it doesn’t comply with its terms of reference for some of the Pentagon’s most-used items on Friday, the source said.
“Anthropic has until 5:01pm on Friday to get on board with the Department of Defense,” a senior Pentagon official said in a statement provided to NBC News, responding to questions about the meeting. “If they don’t come in, the Secretary of War will ensure that the Defense Production Act is applied to Anthropic, forcing them to be used by the Pentagon.”
“Additionally, the Secretary of War will also name Anthropic a supply chain risk,” the official said.
Asked about Tuesday’s meeting, a spokesperson for Anthropic said in a statement: “Dario expressed appreciation for the Department’s work and thanked the Secretary for his work.”
Hegseth praised Anthropic products and said the Pentagon wanted to work with Anthropic, according to another person familiar with the meeting, who asked not to be identified to speak freely. A person confirmed that the Department said it will end the work of Anthropic and the Pentagon on Friday if you do not agree to its terms.
According to reports by The Wall Street Journal and Axios, Anthropic Claude systems were used during the operation to capture Venezuelan President Nicolás Maduro in January. It is not clear exactly how the systems were used.
Hegseth sent a memo to top Pentagon officials Jan. 9 announcing the Pentagon’s initiative toward “the first AI warfare force.” He expressed the drive to use AI models, such as Anthropic, for all legitimate military purposes, “without policy application restrictions” imposed by individual AI companies.
Anthropic is the only AI company whose products are continuously deployed in decentralized networks, through its contract with Palantir, a data analytics company. A senior Pentagon official confirmed to NBC News that xAI reached an agreement with the Pentagon on Monday to use its Grok chatbot system on classified networks, agreeing to allow its systems to be used for “any legitimate use” as Hegseth desired.
Anthropic was one of four AI companies — the others were OpenAI, Google DeepMind and xAI — to receive contracts of up to $200 million in July to “prototype AI capabilities at the frontier of AI that advance US national security.”
