OpenAI, Authors Battle Over Execs' Texts And Proof Of Harm