This is a Paul Ford opinion piece framed around a June 2026 incident where the Trump administration briefly banned Anthropic's Claude Fable 5 model for non-U.S. users citing safety concerns. Anthropic pulled access entirely (reportedly given 90 minutes to comply), then the government lifted restrictions after Anthropic took unspecified steps to address safety risks, and the model returned July 1. Ford's argument appears to center on the "code is free speech" angle—the title suggests he's making a First Amendment claim about AI model restrictions, though the source text cuts off before his actual reasoning. The piece positions this as revealing something about free speech doctrine in America, but the excerpt doesn't include the substantive argument or what "steps Anthropic had taken" actually were. So we have a policy incident and a claimed doctrinal implication, but not the connective tissue explaining why code regulation raises speech issues distinct from other technology controls.