The latest spat between Anthropic and the government raises a broad concern among AI and safety researchers: There is no consistent framework for regulating AI.