Cybersecurity Policy Report, Concerns About Sabotage Led to Anthropic’s Security Risk Designation, Court Told, (Mar 19, 2026)
Concerns that Anthropic might try to sabotage federal government systems led the Trump administration to cancel Anthropic’s contract with the Department of Defense and designate the artificial intelligence company as a security risk, according to court filings.
Amid a contract dispute between Anthropic and DoD, the department became worried that “an AI model used in national security systems would be sabotaged by a hostile and untrustworthy corporate owner,” the administration told the U.S. District Court for the Northern District of California.
Anthropic has asked the court for a temporary restraining order to reverse the penalties imposed on it by the government after the company insisted that DoD not use its AI technology for mass domestic surveillance or autonomous weapons (CPR, March 16).
In its response to Anthropic’s lawsuit, the Trump administration said it had the authority to ban the company from receiving government contracts and did so to protect government operations (Anthropic PBC v. U.S. Department of War, et al., No. 3:26-cv-01996 (N.D. Cal. Mar. 9, 2026)).
“Early this year, the Secretary of War directed the Department of War (‘DoW’ or ‘Department’) to incorporate into any AI service contracts a provision that permits DoW to use the technology for any lawful purpose,” the government noted.
“If it were any other way, an AI provider might gain influence over how DoW conducts operations and which missions it chooses. Anthropic, whose Claude AI model is embedded in DoW’s unclassified and classified systems, refused to accept the term during new contract negotiations, citing its Usage Policy that imposes various limitations on the use of Claude,” it told the court.
“And throughout the negotiations, Anthropic’s behavior more generally caused the Department to question whether Anthropic represented a trusted partner with whom the Department was willing to contract in this highly sensitive area,” it added.
“AI systems are acutely vulnerable to manipulation, and Anthropic could attempt to disable its technology or preemptively alter the behavior of its model either before or during ongoing warfighting operations, if Anthropic—in its discretion—feels that its corporate ‘red lines’ are being crossed. DoW deemed that an unacceptable risk to national security,” the government said.
District Judge Rita Lin has scheduled a March 24 hearing on Anthropic’s request for a restraining order.
News: FederalLegislation LitigationEnforcement DataPrivacy AINews