In a late Friday court filing, AI company Anthropic has formally challenged the Pentagon's assertion that its technology poses an "unacceptable risk to national security," arguing the government's case is built on technical misunderstandings and claims never raised during months of prior negotiations. The company submitted two sworn declarations to a California federal court, directly contesting the Department of Defense's position and revealing that, just a week after the Trump administration publicly declared the relationship over, Pentagon officials told Anthropic the two sides were "nearly aligned."

The legal pushback comes as part of an ongoing dispute between the AI lab and the U.S. government, which has sought to restrict or scrutinize Anthropic's operations on national security grounds. The core of Anthropic's argument rests on the claim that the Pentagon's case relies on a flawed technical assessment of its AI models and their capabilities. According to the filings, the government's allegations were not presented during the extensive negotiation period that preceded the public breakdown and legal action.

This revelation of private alignment contrasts sharply with the public narrative of a complete rupture. The declaration that the parties were "nearly aligned" a week after the Trump administration's pronouncement suggests the negotiations were far more advanced and potentially salvageable than publicly acknowledged. Anthropic's legal team is using this discrepancy to paint the government's subsequent "unacceptable risk" characterization as a sudden and unfounded escalation, not a conclusion drawn from the negotiation record.

The technical nature of the dispute highlights the growing challenges of regulating advanced AI systems within existing legal and security frameworks. Anthropic's defense implies that the Pentagon may be applying broad, traditional national security risk assessments to a novel technology it does not fully comprehend, leading to what the company characterizes as misunderstandings. This case is becoming a test of how specific and technically grounded the government must be when making high-stakes national security claims against AI developers.

The outcome of this legal challenge could set a significant precedent for how the U.S. government interacts with and regulates leading AI companies. If Anthropic succeeds in demonstrating that the Pentagon's risk assessment is technically inaccurate or based on claims not vetted during good-faith negotiations, it could force a more transparent and evidence-based process for future security reviews. Conversely, a ruling in the government's favor would solidify its broad authority to restrict AI development on national security grounds, even in the face of technical counter-arguments from the creators themselves. The case underscores the fragile and often opaque relationship between cutting-edge AI labs and national security agencies, a dynamic that will only grow more complex as the technology advances.