A clash over AI safety has triggered a sharp policy fight between Anthropic, OpenAI, and the U.S. government.
