Anthropic has escalated its legal confrontation with the United States Department of Defense (DoD), submitting two sworn declarations to a federal court in California that challenge the Pentagon’s claim that the company poses an “unacceptable risk to national security.”

The filings, ahead of a key hearing on March 24 before Judge Rita Lin in San Francisco, argue that the government’s position is rooted in technical misunderstandings and allegations that were never raised during prior negotiations.

The dispute follows a late February decision by then-President Donald Trump and Defense Secretary Pete Hegseth to sever ties with Anthropic after the company declined to permit unrestricted military use of its AI systems.

Anthropic’s rebuttal is led by senior officials including Sarah Heck, Head of Policy and a former National Security Council official, and Thiyagu Ramasamy, Head of Public Sector.

In her sworn declaration, Heck—who directly participated in discussions with Defense Secretary Hegseth and Under Secretary Emil Michael—firmly rejected claims that the company sought oversight or approval authority over military operations.

She stated unequivocally that no such demand was ever made during negotiations.

Heck also emphasized that concerns cited by the Pentagon, including fears that Anthropic could alter or disable its technology during military use, were never previously discussed and only surfaced in court filings.

According to her declaration, on March 4—just one day after the Pentagon formally designated Anthropic as a supply-chain risk—Michael emailed CEO Dario Amodei suggesting both parties were “very close” to agreement on contentious issues such as autonomous weapons and domestic surveillance.

This communication appears to contradict subsequent public statements portraying Anthropic as a national security threat, reinforcing the company’s argument that the government’s case lacks consistency and was constructed after negotiations had effectively advanced.