What the court decided
A United States federal judge has struck down the Department of Defense's decision to label the artificial intelligence (AI) company Anthropic a "supply chain risk," ruling on August 27 that the designation was both unconstitutional and unlawful. US District Judge Rita Lin found that the Pentagon's action violated the First Amendment by retaliating against Anthropic for its public criticism of the administration, and violated the Fifth Amendment by denying the company the process it was legally owed before the designation was imposed.
Judge Lin did not mince words about the government's motive. She wrote that the designation was "based on a desire to make a public example out of Anthropic for its 'arrogance'" in criticizing the government, rather than on any real evidence the company posed a security threat. In a separate line, she rejected the government's framing outright: "The empty invocation of national security is not a blank check to punish and retaliate against government critics."
Why the Pentagon targeted Anthropic in the first place
The dispute traces back to Anthropic's refusal to loosen the safety limits it places on its Claude models. Specifically, Anthropic would not lift restrictions that prevent Claude from being used to help build fully autonomous weapons or to conduct mass surveillance of American citizens. Earlier in 2026, Defense Secretary Pete Hegseth and President Trump responded by formally designating Anthropic a supply-chain risk, a label that, once applied, effectively ordered every federal agency, not just the Pentagon, to stop doing business with the company. Court filings note that in the days before the designation, Hegseth had also floated invoking the Defense Production Act against Anthropic, a law that grants the government emergency powers over private industry.
Anthropic responded by filing two separate lawsuits in March 2026, one in California and one in Washington, DC. The ruling issued on August 27 resolves only the California case; the Washington, DC case is still pending, and legal observers expect the government to appeal, potentially all the way to the Supreme Court. Any appeal is likely to turn on a narrower and more consequential question than this case alone: whether courts have the authority to review a national-security designation like this at all, or whether that judgment is entirely the executive branch's to make.
An unusual response from Anthropic
Despite winning, Anthropic's public reaction was notably restrained. A company spokesperson said only that Anthropic remains "focused on working productively with the government to harness AI for our national security," a statement more conciliatory than confrontational given the company had just prevailed in a First Amendment retaliation claim against its own government.
That restraint lines up with a detail buried in the court record: even while the supply-chain designation was in effect, the government was reportedly still in discussions with Anthropic about its Mythos model, a signal that the practical business relationship between Anthropic and federal agencies had not fully collapsed even as the legal fight played out.
Outside legal experts see the ruling as significant for reasons beyond Anthropic specifically. Mark Rasch, a former federal prosecutor now with the security firm Unit221B, framed the core issue this way: "It's one thing for the government to say 'We're not going to do business with you.' It's quite another to say 'Nobody we do business with can do business with you either.'" That distinction, between the government simply choosing not to buy a company's product and the government blacklisting that company across the entire federal supply chain, is what made the designation vulnerable to a due-process challenge.
What it means in practice
For Anthropic, the immediate effect is that federal agencies outside the Pentagon are no longer barred from using Claude, at least while the ruling stands. That matters commercially: government contracts are a significant and growing revenue line for frontier AI labs, and a supply-chain risk label functions as a blacklist across the entire federal government, not just the agency that issued it.
For the wider AI industry, the ruling sets an early marker on how far the government can go in pressuring AI companies to loosen safety commitments through contracting leverage rather than legislation or formal rulemaking. Every major AI lab now sells to, or wants to sell to, the federal government, and each has its own internal limits on military and surveillance use. This case is the first time a court has weighed in on whether the government can effectively punish a company for holding the line on those limits. Given that a second lawsuit is still active and an appeal is expected, this ruling is unlikely to be the final word.
