How Anthropic is pursuing a state-by-state push for ever-tougher AI safety laws, in contrast with OpenAI's "reverse federalism" strategy for common state rules (Politico)