The back and forth between Anthropic and the U.S. government highlights broader tensions over AI safety, sovereignty and vendor control in defense applications.
