District Judge Rita Lin suggested that the government’s evidence failed to justify the blacklisting and noted that the record appeared to have weakened since the legal challenge began.
The judge specifically questioned the government's claims, stating she saw no proof that Anthropic could manipulate its models or implement a "kill switch" after delivery to the military.
The legal dispute originated earlier this year after Anthropic and the Department of Defense clashed over the terms of use for the company's Claude AI models.
While the military sought unrestricted access for all applications, Anthropic insisted on prohibitions regarding mass surveillance and fully autonomous weaponry.
This standoff led the Pentagon to label the firm a risk, a move the judge previously characterized as a potential attempt to cripple the company.
Both Anthropic and the Pentagon have requested a summary judgment to resolve the case.
While Judge Lin has already issued a temporary injunction blocking the ban, a final written decision is expected in the coming months.
The outcome of this case is seen as a critical milestone in defining the relationship between the U.S.
government and the AI companies currently helping to shape federal safety regulations.