District Judge Rita Lin, appointed by former U.S. President Joe Biden, has temporarily blocked the Pentagon’s “ban” on Anthropic, which classified the company as a “supply chain risk” after the latter refused to remove restrictions on the use of its artificial intelligence systems.
THE BACKGROUND
Known primarily for the Claude chatbot and funded by Google and Amazon, Anthropic was founded in 2021 in California by a group of former OpenAI employees, including siblings Dario Amodei (the CEO) and Daniela Amodei (the president). To try to distinguish itself from the competition, the startup focused on user data protection and the development of “ethical” artificial intelligence systems, meaning those that are not harmful to society.
For about a month, a political and legal clash has been underway in the United States between Anthropic and the Department of Defense (also known as the Pentagon) regarding the use of the startup’s technologies for military purposes. The American government wants to have unrestricted access to the Claude language model; however, Anthropic opposes its use for mass surveillance or lethal military actions carried out with autonomous weapons.
Following Anthropic’s refusal to remove operational constraints, the Pentagon classified it as a “supply chain risk,” a designation applied to entities that risk exposing military systems to infiltration or sabotage by enemies, which would have the effect of excluding Anthropic from public contracts, as well as inflicting severe economic and reputational damage. This was the first time the U.S. government classified an American company as a supply chain risk.
In response, Anthropic sued the U.S. government, accusing it of violating the constitutional right to freedom of expression and arguing that the Pentagon’s measure is retaliation for its ethical stance.
JUDGE LIN’S DECISION
According to Judge Lin, Anthropic is right: the ruling will take effect in seven days, however, so that the Trump administration has time to file an appeal.
In the ruling, Lin states that the actions of the Trump administration do not appear motivated by national security concerns, as claimed, but rather by a desire to punish Anthropic. “The acts,” the judge wrote, “support the hypothesis that Anthropic is being punished for criticizing the government’s procurement position in the press. Punishing Anthropic […] constitutes classic illegal retaliation under the First Amendment” of the Constitution.
The government’s position, instead, is that the supply chain risk designation derives from Anthropic’s refusal to accept contractual conditions.
WHAT ANTHROPIC SAID
Anthropic, through spokesperson Danielle Cohen, expressed satisfaction with the judge’s decision while emphasizing that its goal remains “to work constructively with the government to ensure that all Americans can benefit from safe and reliable artificial intelligence.”
ARESUS’S COMMENT
“Anthropic achieves a first important legal victory against the Pentagon. The use and abuse of ‘national security’ powers by the government continues to have constraints and limits within the U.S. judicial system,” wrote on X Alessandro Aresu, analyst and author of Geopolitics of Artificial Intelligence.
