The U.S. federal court system has issued a landmark ruling that could redefine the rules of engagement between the state and artificial intelligence developers. Judge Rita Lin ruled that the Pentagon's actions toward Anthropic were not merely erroneous but directly contrary to the U.S. Constitution. The "supply chain risk" status imposed on the developer of the Claude model has been vacated, and all related restrictions have been immediately suspended.
The key aspect of this verdict is the recognition that the conflict had nothing to do with technical reliability or Anthropic's ability to fulfill government contracts. The dispute centered on the company's public stance regarding permissible use cases for AI. Judge Lin saw in the Pentagon's actions a direct attempt to punish the company for its views, which constitutes a blatant violation of the First Amendment.
Moreover, the procedure was also deemed flawed from a procedural standpoint. Anthropic was deprived of the opportunity to review the allegations in advance and contest them, violating the due process guarantees enshrined in the Fifth Amendment. The ruling took effect immediately, although the administration of President Donald Trump retains the right to appeal.
The Root of the Conflict
The cause of the standoff was the terms of use for Anthropic's models by the military. The company permitted the use of Claude in the realm of national security but categorically refused to lift bans on mass surveillance of American citizens and the development of fully autonomous weapons systems. As early as February, the Pentagon issued an ultimatum, demanding the removal of all restrictions "for any lawful purposes," threatening blacklisting. After CEO Dario Amodei refused, the threat was carried out, and soon followed Trump's order to cease the use of the company's technologies by all federal agencies within six months.
The conflict sent shockwaves through the entire industry. Sam Altman, head of OpenAI, publicly urged the military not to blacklist Anthropic, and later acknowledged that certain aspects of his own deal with the Pentagon were a mistake. Notably, despite the restrictions, Anthropic's technologies continued to be used by individual agencies—in April, the U.S. National Security Agency used the Mythos model.
My analysis: This ruling creates a dangerous precedent for the state, but a crucial one for the industry. It clearly signals that administrative pressure on companies for their ethical stance is unacceptable. However, Anthropic's victory may prove Pyrrhic—the government will likely find new legal mechanisms for pressure, now taking judicial practice into account. The question of who controls the boundaries of AI use in the military domain remains open, and this verdict is merely the first round in a protracted battle.