US judge overturns Pentagon blacklist of Anthropic
Developing story first seen 2 hours ago
A federal judge has formally overturned the Trump administration’s blacklisting of Anthropic, finding that the Pentagon’s designation of the AI company as a national-security “supply-chain risk” was unconstitutional retaliation. The ruling removes restrictions imposed across nine federal agencies and allows Anthropic, its military contractors and suppliers to resume doing business with the government, though the Pentagon remains free to select competing AI systems.
US District Judge Rita Lin vacated Defence Secretary Pete Hegseth’s 27 February decision and a separate ban affecting firms linked to the US military, calling the measures unlawful and baseless. They followed the collapse of negotiations over a $200 million deal for Anthropic’s Claude models, after the company sought limits on use in lethal autonomous weapons and mass surveillance while the Pentagon argued for “all lawful use”; the department is expected to appeal, and a separate Anthropic case in Washington DC continues.
- Judge overturns Pentagon’s Anthropic blacklist as unconstitutional retaliation.
- Federal contracting and supplier restrictions on Anthropic have been lifted.
- Dispute centred on military AI-use limits under a $200 million deal.
New here? Start with this
Anthropic is a US artificial-intelligence company best known for Claude, a system that can analyse information and generate text. The Pentagon has been exploring such systems for tasks including research, planning and handling large amounts of data, and had agreed a contract with Anthropic worth up to $200 million.
The disagreement centred on how the military could use Anthropic’s technology. Anthropic wanted restrictions on use in lethal weapons that operate without direct human control and in large-scale surveillance, while the Defence Department said it needed access for any use permitted under US law.
A national-security supply-chain designation is normally used when the government believes a company’s products or services could pose a security risk. Such a label can limit a company’s access to federal contracts and affect firms that work with it, making it significant for both the government’s technology procurement and the wider AI industry.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Supporters of the ruling argue that the government must not use national-security designations to punish a company for expressing principled limits on how its technology may be used. They contend that Anthropic can reasonably refuse involvement in lethal autonomous weapons and mass surveillance while still competing for public contracts, and that supply-chain powers require genuine evidence of risk rather than serving as leverage in a contractual dispute. This view prioritises constitutional protections, due process and safeguards against the politicisation of national-security authorities.
The case against
Critics argue that the Pentagon needs dependable access to advanced AI and broad discretion to deploy it for any lawful military purpose, particularly when national security and service members’ safety may be at stake. They may say that a contractor seeking major defence work should not be able to impose restrictions that constrain commanders or create uncertainty in urgent operations, and that the department should be free to favour suppliers whose systems meet its operational requirements. This view prioritises military readiness, democratic control of defence policy and the government’s ability to choose technology partners without judicially imposed commercial obligations.
Full account
A US federal judge has overturned the Pentagon’s designation of Anthropic as a national-security-related supply-chain risk, finding that the restrictions imposed on the artificial intelligence company were unlawful. US District Judge Rita Lin ruled that the government had used sweeping measures against Anthropic in a manner that amounted to unconstitutional retaliation, rather than a justified response to a genuine security threat.
The dispute followed a breakdown in talks over the military use of Anthropic’s Claude models. Anthropic sought limits intended to prevent its systems being used for mass surveillance or lethal autonomous weapons, while defence secretary Pete Hegseth maintained that the government should be able to make any lawful use of technology obtained under contract. After negotiations failed, the administration moved to bar Anthropic from federal contracting and to restrict military contractors and suppliers from dealing with it.
Lin’s 59-page decision vacated the February designation and blocked enforcement of associated measures by several federal bodies. She said references to national security could not be used as a general licence to penalise a company for criticising the administration’s position. The ruling does not oblige the Pentagon to resume work with Anthropic or to select its models over rival products. Anthropic welcomed the judgment and said it remained willing to work with government on national-security applications.
The case is part of a wider confrontation between Anthropic and the Trump administration. Reports say the company brought separate legal challenges concerning government action against it, meaning aspects of its status may still depend on another court decision. The Pentagon was expected to consider an appeal, while the company’s underlying disagreement with the government over safeguards and permitted military applications remains unresolved.
Where outlets differ
Source 1 presents the judgment as removing sanctions imposed by nine agencies and says the Pentagon’s restrictions on contractors and suppliers were set aside. Source 2 describes the ruling as applying to certain agencies and says Anthropic may still be regarded as a supply-chain risk pending a separate case.
Source 1 places greater emphasis on the reported use of Claude during an operation involving Venezuelan president Nicolás Maduro and on the subsequent internal concerns relayed to officials. Source 2 instead focuses on President Trump’s reported six-month federal phase-out order and Anthropic’s separate lawsuits.
Source 2 adds business context about Anthropic’s reported revenue growth and a possible initial public offering; Source 1 concentrates on the legal reasoning and the contract dispute.
More coverage
- The Hill — Judge rules Pentagon's supply chain risk designation for Anthropic was illegal
- NPR — Judge says Pentagon's measures against Anthropic were 'illegal and baseless'
Read the full article at the source →
Originally published by Wired as “A Judge Has Blocked the Pentagon’s Attempt to Blacklist Anthropic”.