OpenAI and Google claim that restricting AI model training on copyrighted data would give China a competitive edge in AI development.