OpenAI and Google have asked for government permission to train their AI models using copyrighted materials