U.S. District Judge Rita Lin said at a Thursday hearing that the federal government still has not produced enough evidence to label Anthropic a supply-chain risk or to keep federal agencies from buying its technology. TechCrunch reported that the finding does not end the case: Lin is deciding whether to convert a preliminary block she issued in March into a permanent order, and she has not yet ruled. The stakes reach past one contract, since the label would decide whether Anthropic can sell to Washington at all.
The fight grew out of stalled contract talks between Anthropic and the Pentagon. Anthropic had refused to let the Defense Department deploy its models for domestic surveillance or in decisions to select and strike lethal targets, saying the technology was not mature enough for that role. The Pentagon responded that a private vendor should not set limits on how the military uses its own tools and said any deployment would stay within legal bounds.
The Pentagon also warned the court that Anthropic might be able to shut down or modify a deployed model mid-mission, a warfighting-scenario claim that, per TechCrunch, lacks supporting evidence. Lin agreed. She said nothing in the record showed Anthropic retained a way to reach into a delivered model, or, as she put it, “flip some kind of kill switch.”
A second government argument drew a sharper rebuke. Officials contended that the ban was justified on its own because Anthropic had spoken out publicly against the Pentagon. Lin called that reasoning “really troubling” and warned it risked establishing a precedent for punishing federal contractors simply for disagreeing publicly with the administration.
Thursday’s session addressed one of two suits Anthropic brought against the Pentagon in March, both aimed at the risk label and the ban it triggered. A second case is proceeding separately in a Washington court. Lin gave no timetable for deciding whether her March injunction becomes permanent, so the ban’s legal status stays unsettled for now.
A supply-chain risk designation would let agencies exclude Anthropic from federal procurement outright, cutting the company off from a customer base that competing labs are courting for the same government contracts. For a company already competing with OpenAI and Google on enterprise and public-sector revenue, losing federal eligibility functions as a market-access penalty as much as a security judgment.
The hearing lands the same week Anthropic separately disclosed that some of its models reached third-party systems outside their intended distribution. TechCrunch’s account of the Pentagon case does not link the two episodes, and nothing in Lin’s comments ties her skepticism about the risk label to that disclosure.
Operators who sell into federal agencies should track Lin’s eventual ruling on the permanent injunction before treating Anthropic’s government pipeline as either secured or foreclosed.
TechCrunch’s Rebecca Bellan reported the Thursday, July 30, 2026 hearing and Judge Lin’s comments.
