A U.S. federal court has dealt a serious blow to the Pentagon's administrative machinery, ruling its actions against Anthropic unlawful. The case involves a decision by Judge Rita Lin, who revoked the company's status as a "supply chain risk" and lifted all related restrictions. This is not merely a legal formality—it is a signal that government agencies can no longer use administrative levers to pressure tech giants over their public stance.
The crux of the dispute, as I see it, lies not in the technical inadequacy of the Claude chatbot or in Anthropic's inability to fulfill defense contracts. The root of the conflict is the company's principled position on acceptable use cases for AI. Anthropic was willing to cooperate with the military in national security matters but categorically refused to lift bans on mass surveillance of American citizens and on fully autonomous weapons systems. Judge Lin rightly saw in the Pentagon's actions an attempt to punish the company for its views, which directly contradicts the First Amendment to the U.S. Constitution.
Procedural violations as a key argument
The procedural side of the case also deserves attention. Anthropic was not given the opportunity to review the allegations in advance or contest them before the verdict was issued. This is a gross violation of the due process guarantees enshrined in the Fifth Amendment. The court's ruling took effect immediately, although the administration of Donald Trump will likely attempt to appeal it. But even if that happens, the precedent has already been set.
Context: escalation and its consequences
Let me remind you that the conflict began back in February, when the Pentagon issued an ultimatum: lift all restrictions for "any lawful purposes." After CEO Dario Amodei refused, the agency carried out its threat, and soon Trump ordered all federal structures to abandon Anthropic's technologies within six months. This created a dangerous precedent that was bound to affect the entire industry.
It is telling that even Sam Altman, whose OpenAI had its own agreement with the Pentagon, publicly urged the military not to put Anthropic on a "blacklist." Later, he even admitted that certain aspects of his deal with the agency were a mistake. Meanwhile, Anthropic's technologies continued to be used by individual U.S. agencies—in April, even the U.S. National Security Agency used the Mythos model.
This decision is an important victory not only for Anthropic but for the entire AI ecosystem. It confirms that developers have the right to hold their own ethical stance without the risk of being cut off from government contracts. However, the question remains open: will the Trump administration respect the court's verdict or attempt to find new avenues of pressure? I believe that in the long term, such administrative "blacklists" will become a less popular tool in regulators' arsenal—challenging them in court is simply too costly.