The legal question should be around: Does open AI make any guarantees about the content and their validity? Is the output considered property of OpenAI? You can

The legal question should be around: Does open AI make any guarantees about the content and their validity? Is the output considered property of OpenAI? You can’t train on it, so does that mean it’s their property? This is the most important question. For example, if you were to hire a group of writers, ghost writers to be specific, to re-create someone else’s work, then publish it yourself as your own who is at fault? Who gets sued? Is it you who directed the writers to do the work or is it the writers who did the work? So if we accept this model, the owner/onus is on the prompter to not re-create copyrighted works. (Albeit an assumption that the writers are unaware to some extent)