📰 Key Summary

Anthropic won a legal battle against the US federal government on Thursday night. California federal district judge Rita Lin ruled that the Trump administration’s move to label Anthropic a “supply chain risk” was unlawful. In her ruling, she found that Defense Secretary Pete Hegseth’s designation of Anthropic as a national security threat amounted to “unlawful retaliation” in violation of the First Amendment, and that the decision was “arbitrary and capricious”; Anthropic was also denied the due process protections required under the Fifth Amendment. The case stems from Hegseth and Trump labeling Anthropic a supply chain risk earlier this year and ordering all federal agencies — even ones unrelated to defense — to stop working with the Claude developer. At the heart of the dispute is Anthropic’s hard line on safety safeguards: the company refused to let the Pentagon use its models for fully autonomous weapons or mass surveillance of American citizens, while the Defense Department denies any such use and instead accuses Anthropic of trying to control the military’s use of models it paid to procure. In her ruling, Judge Lin wrote that the government’s “words and actions confirm that these measures were motivated by a desire to publicly punish Anthropic” for criticizing the government as “arrogant,” and pointed out a contradiction: Hegseth on one hand proposed invoking the Defense Production Act to compel Anthropic’s cooperation (implying the company is critical to national security rather than a threat to it), while on the other hand the Defense Department kept negotiating contracts with the company, and the government is currently working with Anthropic on a cybersecurity project involving its new model, Mythos. Lin also emphasized that after Anthropic handed its technology over to the Defense Department, it was “confirmed” that no backdoor access existed. She wrote that while the Defense Department is free to choose its AI vendors, the sweeping restrictions imposed on Anthropic were “unlawful and baseless,” and that “a bare invocation of national security is not a free pass to punish and retaliate against critics of the government.” An Anthropic spokesperson welcomed the ruling, saying the company will keep working with the government to put AI to use for national security in ways that benefit all Americans. Anthropic filed suit against the Defense Department in both California and Washington, D.C. back in March; the D.C. case is still pending.

💬 JudyAI Lab Take

Anthropic beat the Trump administration in court on Thursday, with California federal judge Rita Lin ruling that Defense Secretary Hegseth’s “supply chain risk” designation was unconstitutional — unlawful retaliation against a government critic.

The core tension in this case is worth flagging for AI builders: the Defense Department wanted to invoke the Defense Production Act to force Anthropic’s cooperation on one hand, while slapping a blanket ban on the company with the other — a contradiction that exposes the growing friction between “safety red lines” and “government procurement leverage.” Anthropic held firm on refusing to let its models be used for fully autonomous weapons and mass surveillance, and that refusal itself apparently got read by the government as “arrogance” worthy of retaliation. The judge specifically called out that a bare invocation of national security can’t be a free pass to punish critics — that’s a key precedent for any AI company doing business with enterprises or governments: setting usage restrictions is a legitimate right for a technology provider, not something that should be twisted into hostility.

For readers, this is a good reminder to pay attention to a vendor’s publicly disclosed usage policies and red lines when evaluating AI providers — that often tells you more about a company’s actual values than its marketing copy ever will.

📅 Original Article Info

🔗 Further Reading