ع
US District Judge Rita Lin determined that the Defense Department’s February 2026 designation of Anthropic as a supply chain risk amounted to unconstitutional retaliation for the company’s protected speech on AI safety limits, according to her 59-page ruling issued the previous day. The order permanently enjoins multiple federal agencies from enforcing the blacklist, which had barred Pentagon contractors from doing business with the AI startup, while declaring the actions violated both the First Amendment and the Due Process Clause of the Fifth Amendment. Lin wrote that the empty invocation of national security is not a blank check to punish and retaliate against government critics, adding that the measures were arbitrary, capricious and illegal.
The dispute originated after Anthropic declined to revise its usage policies to allow its Claude models for all lawful military purposes, specifically rejecting applications involving domestic mass surveillance of Americans or fully autonomous lethal weapons systems where human oversight was absent. Defense Secretary Pete Hegseth labeled the company a supply chain risk on February 27, 2026, a step previously reserved for entities tied to foreign adversaries under an obscure procurement statute, and the White House described Anthropic as a radical left, woke company attempting to dictate terms to the military. The Pentagon maintained at the time that private firms should not constrain military operations, while a Congressional Research Service report from April detailed how the clash raised questions for lawmakers over balancing AI innovation with defense needs.
Anthropic filed suit in March arguing that the designation was unprecedented and unlawful, claiming it had already secured government contracts worth up to $200 million as part of a July 2025 Defense Department initiative that also included awards to OpenAI, Google and xAI. In earlier proceedings Lin had temporarily blocked the blacklist, stating the government appeared intent on crippling the company and chilling public debate over military AI applications. The final ruling found no evidence that Anthropic posed an actual security threat or would sabotage its technology, concluding instead that the moves aimed to make a public example of the firm for what officials called its arrogance in criticizing the government.
The company welcomed the decision in a statement, saying it remained focused on working productively with the government to harness AI for national security so all Americans benefit from this technology. Anthropic had maintained that its models were not reliable enough for safe deployment in weapons systems and that surveillance uses would violate rights, positions that drew support from some former national security officials who filed briefs arguing the actions reflected pretextual punishment. The ruling does not compel the Pentagon to contract with Anthropic and leaves the department free to select other vendors through standard processes.
OpenAI reached its own agreement with the Defense Department shortly after the Anthropic designation, accepting terms for all lawful purposes while implementing technical guardrails to align with its red lines against mass surveillance and autonomous lethal weapons, the company said in a February 2026 statement. That deal followed reports of Claude models being used in a January operation to capture Venezuelan President Nicolás Maduro, which heightened Pentagon concerns about potential company veto power over deployments. According to a Library of Congress analysis, the episode underscored broader tensions in AI defense contracting, where multiple firms hold contracts valued at up to $200 million each to accelerate adoption for national security challenges.
The decision takes effect immediately although the government is expected to appeal, with Lin retaining jurisdiction to enforce the order against the named agencies including the Pentagon, State Department and Department of Homeland Security. No Pentagon spokesperson commented on the outcome by late Thursday, consistent with the lack of immediate response following the BBC’s outreach to both the department and the White House. The case marks the first instance in which a domestic AI company faced such a public national security designation, potentially influencing how other technology providers approach future military collaborations.
