US Regulators Direct Anthropic to Halt Distribution of AI Models