Anthropic Sues to Reverse Pentagon ‘Supply-Chain Risk’ Blacklist After AI Safeguard Talks Collapse

Big civilian tech firms squaring off with the U.S. security state is not new: in 1999-2000, Sun Microsystems and RSA sued to strike down the Clinton administration’s export controls on strong encryption, framing code as protected speech—an argument the Ninth Circuit largely upheld in Bernstein v. U.S. (2003). Today’s AI clash reprises that tension but on a larger canvas: algorithms, not ciphers, now underpin everything from drone targeting to corporate productivity. The Pentagon’s unprecedented use of a ‘supply-chain risk’ statute—previously aimed at Huawei (2019) and Kaspersky (2017)—against a domestic firm signals a shift from foreign-centric security fears to disciplining home-grown platforms that resist unconstrained military integration. Over the next century, the result could shape whether frontier AI evolves under a militarized, state-directed paradigm (à la nuclear physics post-1945) or under a pluralistic regime where private actors assert normative limits. Whichever side prevails, this moment will be remembered as an early test of who sets the red lines for thinking machines: elected governments or the engineers who build them.