“Anthropic never had the ability to stop the cloud from working, alter its functionality, shut down access, or otherwise impact or jeopardize military operations,” wrote Thiagu Ramasamy, Anthropic’s head of public sector. “Anthropic does not have the necessary access to disable the technology or change the behavior of the model before or during ongoing operations.”
The Pentagon has been sparring with the leading AI lab for months over how its technology can be used for national security — and what the limits on that use should be. This month, Defense Secretary Pete Hegseth labeled Anthropic a supply-chain risk, a designation that would bar the Defense Department, including contractors, from using the company’s software in the coming months. Other federal agencies are also abandoning the cloud.
Anthropic filed two lawsuits challenging the constitutionality of the ban and is seeking an emergency order to overturn it. However, customers have already started canceling deals. One of the cases is scheduled to be heard on March 24 in federal district court in San Francisco. The judge may decide on a temporary reversal shortly thereafter.
In a filing earlier this week, government lawyers wrote that the Defense Department “does not need to tolerate the risk that critical military systems will be jeopardized at critical moments for national defense and active military operations.”
The Pentagon is using the cloud to help analyze data, write memos and prepare war plans, WIRED reports. The government’s argument is that Anthropic could disrupt active military operations by cutting off access to the cloud or pushing harmful updates if the company disapproves of certain uses.
Ramasamy rejected that possibility. “Anthropic does not maintain any backdoor or remote ‘kill switch’,” he wrote. “For example, anthropogenic personnel cannot log into DoW systems to modify or disable models during an operation; the technology simply does not work that way.”
He further said that Anthropic would only be able to provide updates with the approval of the government and its cloud provider, in this case Amazon Web Services, though he did not specify this by name. Ramasamy said Anthropic cannot access signals or other data that military users enter into the cloud.
Anthropic officials have said in court filings that the company does not want veto power over military strategic decisions. Policy chief Sarah Heck wrote in a court filing Friday that Anthropic was willing to provide the same guarantees in a contract proposed on March 4. “for the avoidance of doubt, [Anthropic] “It is understood that this license does not grant or confer any authority to control or veto legitimate Department of War operational decision-making,” the proposal said, according to the filing, which refers to an alternative name for the Pentagon.
Heck claimed the company was also willing to accept language that would address his concerns about using the cloud to help carry out deadly attacks without human supervision. But ultimately the talks broke down.
For now, the Defense Department said in court filings that it is “taking additional measures to mitigate the supply chain risk” posed by the company, “working with third-party cloud service providers to ensure that Anthropic leadership cannot unilaterally make changes to cloud systems currently in place.”
<a href