California enacts AI safety evaluation laws backed by Anthropic and OpenAI