Judge blocks Pentagon blacklisting of Anthropic
A U.S. judge temporarily blocked the Pentagon's move to blacklist Anthropic over battlefield‑use AI safety concerns, highlighting tensions between defense procurement and AI governance. The ruling reflects legal and ethical pushback as governments try to restrict access to advanced models for military use. (x.com)
U.S. District Judge Rita F. Lin issued a 43‑page order granting a preliminary injunction on March 26, 2026, in the San Francisco case captioned Anthropic PBC v. Department of Defense. (justia.com) Lin wrote that the record supports an inference the Defense Department acted to punish Anthropic for public criticism, calling that “classic illegal First Amendment retaliation.” (politico.com) The opinion cites internal DOD documents saying Anthropic’s “risk level escalated” because it was “engaging in an increasingly hostile manner through the press.” (defenseone.com / justia.com) Lin’s injunction bars enforcement of President Trump’s governmentwide directive to halt agency use of Anthropic’s Claude models and restrains the Pentagon’s supply‑chain‑risk labeling while the litigation proceeds. (apnews.com / cnbc.com) The judge stayed her order for seven days to give the government time to seek emergency relief from an appeals court, creating a short window for the administration to ask the Ninth Circuit to intervene. (politico.com) Anthropic had entered a two‑year prototype agreement with the DOD’s CDAO in July 2025 with a reported $200 million ceiling and had already deployed Claude variants on some classified networks, which the company says it risks losing without the injunction. (anthropic.com / cnbc.com) Lin emphasized that the supply‑chain‑risk statute at issue has historically targeted foreign adversaries and had not before been applied to a domestic U.S. company, a point she used to question the stated national‑security rationale for the designation. (justia.com / politico.com)
