Defense Department says Anthropic poses ‘unacceptable risk’ to national security

The Division of Protection mentioned giving Anthropic continued entry to its warfighting infrastructure would “introduce unacceptable danger” to its provide chains in a court filing submitted in response to the AI firm’s lawsuit. If you happen to’ll recall, Anthropic sued the government to problem the provision chain danger designation it obtained for refusing to permit its mannequin for use for mass surveillance and the event of autonomous weapons.

In its submitting, the division defined that its secretary, Pete Hegseth, had a provision integrated into AI service contracts, permitting the company to make use of their applied sciences for any lawful objective. Anthropic refused its phrases and apparently, the corporate’s conduct triggered the Pentagon to query whether or not it really was a “trusted accomplice” that it might work with with regards to “extremely delicate” initiatives. “In spite of everything, AI methods are acutely weak to manipulation, and Anthropic might try and disable its know-how or preemptively alter the conduct of its mannequin both earlier than or throughout ongoing warfighting operations, if Anthropic — in its discretion — feels that its company “pink strains” are being crossed,” the Pentagon wrote in its submitting. “DoW deemed that an unacceptable danger to nationwide safety,” it added, referring to the company because the Division of Warfare, which is the Trump administration’s most well-liked identify for it.

It was because of these considerations that President Trump ordered federal agencies to cease utilizing its know-how, the submitting reads. The corporate is asking the courtroom to situation a preliminary injunction and put a pause on a ban whereas it’s difficult its provide chain danger designation in courtroom. Whereas Anthropic’s shoppers might proceed working with the corporate on non-defense-related tasks, it says the label might trigger it to lose billions of {dollars} in income. It’s not fairly clear if Anthropic continues to be making an attempt to reach a new deal with the federal government, as was reported earlier than it filed its lawsuit. As The New York Times notes, Microsoft, Google and OpenAI had filed friend-of-the-court briefs in assist of Anthropic since then.