The U.S. Department of Defense said late on Tuesday local time that the "red lines" set by artificial intelligence company Anthropic on the use of its technology made the company pose an "unacceptable national security risk." This was the department's first official response to a lawsuit filed by the company over its "supply chain risk" determination. This comes after Defense Secretary Pete Hegseth last month classified Anthropic as a "supply chain risk" entity, and Anthropic asked a court to temporarily prohibit the Department of Defense from enforcing that label.
In a 40-page filing in a federal court in California, the Department of Defense stated that its core concern is that if Anthropic believes that military use violates the company's corporate "red lines," the company may "attempt to deactivate its technology or preemptively modify model behavior before or during combat operations." The Department of Defense believes that this uncertainty may directly affect the degree of reliance and controllability of this technology in war operations.
Information shows that Anthropic signed a $200 million contract with the Pentagon last summer, involving the deployment of its artificial intelligence technology in classified systems. During subsequent negotiations on contract terms, Anthropic stated that it did not want its system to be used for mass surveillance of Americans and believed that the technology was not yet suitable for direct use in target selection or firing decisions with lethal weapons. The Pentagon emphasized that private companies should not have the final say on how the military uses technology, and the differences between the two sides on the boundaries of technology application became public.
After the Department of Defense classified Anthropic as a "supply chain risk," it triggered criticism from many parties. Multiple organizations pointed out that the Pentagon could have expressed its position by terminating the contract without having to use the more lethal tool of supply chain risk identification. Employees of many technology companies, including OpenAI, Google and Microsoft, as well as multiple legal rights groups, have submitted "amicus curiae" submissions to the court to publicly support Anthropic's position.
In the lawsuit, Anthropic accuses the Department of Defense of violating its First Amendment rights, arguing that the government is punishing companies based on ideological positions. The company contends that companies have the right to express their views on the ethical boundaries of their technology and set limits accordingly, and should not be labeled a security risk because of differences in values with the government.
Currently, the court has scheduled a hearing on Anthropic’s preliminary injunction application for next Tuesday. As of press time, Anthropic has not publicly responded to the latest Department of Defense document.
Related articles:
Anthropic seeks appeals court to halt Pentagon finding of risks to its supply chain