The severing of relations with Anthropic concludes a week of tense negotiations between the artificial intelligence company and the government, which could have significant implications for the future application of this technology. The Pentagon is eager to use Anthropic's AI system Claude in its closed networks for various legitimate purposes; however, there are two key positions that the company is not willing to change: the system cannot be used for autonomous weaponry and mass surveillance of U.S. citizens.
Defense Secretary Pete Hegseth later noted on platform X that Anthropic will be considered a "supply chain risk," a designation typically applied to companies perceived as potential threats from foreign competitors.
In its statement, Anthropic announced that it will challenge this decision in court. "Intimidation or punishment from the Department of Defense will not affect our stance on mass domestic surveillance or fully autonomous weapons," their comment published on Friday stated.
The conflict escalated during a meeting at the Pentagon between Hegseth and the company's CEO Dario Amodei. Although the meeting was conducted in a friendly atmosphere, Trump's comments over the weekend indicate a deterioration of the situation. Earlier this week, Anthropic stated that it does not plan to concede to the Pentagon's demands.
According to Pentagon Deputy Undersecretary for Research and Development Emil Michael, the deal with Anthropic was "practically finalized" when the company made its statements about not intending to compromise. Pentagon spokesperson Sean Parnell emphasized that they will not allow any company to set the terms for their operations.
On Friday, Trump called Anthropic's actions a "catastrophic mistake" and accused the company of trying to dictate terms for military operations. Shortly after, the General Services Administration announced plans to exclude Anthropic from the list of contractors on the USAi.gov website, which serves as a central platform for AI testing.
"No contractor, supplier, or partner working with the U.S. military will be allowed to do business with Anthropic," Hegseth added.
AI industry representatives have come to Anthropic's defense, including OpenAI CEO Sam Altman, who expressed support for the company's concerns regarding cooperation with the Pentagon.
What work was Anthropic doing with the Pentagon?
Anthropic's AI model Claude became the first to be implemented in secret military networks. Last summer, the company signed a contract with the Pentagon worth up to $200 million. Unlike Anthropic, other major players in the AI field, such as OpenAI, have only worked with unclassified networks.
The Anthropic contract includes an acceptable use policy that prohibits the use of the Claude system for mass surveillance and autonomous weaponry.
"This conflict arises at an inopportune time, as the Department of Defense has been quite positive about Claude and claims that usage restrictions have never been applied in their practice," noted Gregory Allen, a senior advisor at the Center for Strategic and International Studies, on Bloomberg Radio.
Nevertheless, the Pentagon does not wish to be limited by any company's policies. A Pentagon representative clarified that "tactical operations cannot be based on exceptions," and that the responsibility for compliance with the law lies with the military as the end users.
In the Pentagon's view, it is crucial to avoid national security risks that could arise from the need to seek permission from the company.
Severing ties with Anthropic could create problems for the Pentagon, as it will need to find replacements for all systems using Claude. Although a Pentagon representative stated that Elon Musk's AI system Grok is suitable for use in classified conditions, it is not considered as advanced as Claude.
How will this affect Anthropic's business?
For Anthropic, losing the $200 million contract is not a critical threat, given that its market capitalization is around $380 billion. However, there is a supply chain risk, meaning that any company working with the U.S. military will have to confirm that it has no ties to Anthropic.
A significant portion of Anthropic's success is tied to corporate contracts with large companies, many of which may have dealings with the Pentagon.
"This means that a substantial part of Anthropic's client base could disappear due to their government contracts or their desire to secure such contracts in the future," noted Adam Connor, Vice President for Technology Policy at the Center for American Progress.
Jensen Huang, CEO of Nvidia, expressed hope for an agreement between the Pentagon and Anthropic, but added that "if that doesn't happen, it's not the end of the world," as there are other AI companies that the Pentagon can collaborate with.
The Pentagon is also considering invoking the Defense Production Act of 1950, which gives the president broad powers to control domestic industry, to compel Anthropic to cooperate. It is unclear whether the Pentagon can simultaneously recognize Anthropic as a supply chain threat and force it to comply with its terms.
Trump's statement did not indicate whether a plea agreement is anticipated.
"Anthropic is not the only company under threat," Connor emphasized. "The Pentagon's actions serve as a warning to other AI companies that wish to profit from government collaboration." Connor also added that this sends a signal to other AI companies that they should not attempt to impose restrictions on the use of their technologies.
If the Pentagon simply disapproved of Anthropic's terms, it could easily terminate the contract and turn to another company; however, experts believe that the government is eager to continue using Anthropic's technologies and is applying all possible leverage to achieve this.
It remains unclear how the military will replace Anthropic's systems and what steps the administration will take in the future.
"In the context of an AI race with China, where the White House compares it to the Cold War space race, you wouldn't want to destroy one of the most promising companies in the industry," Allen concluded on Bloomberg.
"There is a more sensible way to resolve this conflict than the hardline stance taken by the administration."
