> Any
> decision ruling that as derivative work requiring licenses would also kill
> Wikipedia, Goodreads, IMDB and hundreds of other places on the Internet
> that describe and books, tv shows or movies in any kind of detai.
>...
That's not true.
The jury could for example decide that summaries do not have a negative
impact on the market of the copyrighted work, but that generated sequels
have a negative impact on the market for the upcoming "The Winds of Winter".
I was refering specifically to summaries.
The sequels are an intersting question in itself. In the context of the lawsuit itself they are as laughable as the summaries as they never shown that the LLMs were capable of generating full texts of the sequels in question in quality sufficient to *actually* compete on any type of marketplace. They only submitted one-page *outlines* of a supposed sequel. Which is ..... a summary again. No *actual* enfringement, beyond fair use was demonstrated. I would be *very* surprised if the lawyers would be able to manage to create a full lenght, detailed and coherent narrative book that could fool anyone in the jury that it is a real book in the GoT series. And without that, the lawsuit is meriless. A one-page summary is not competing in the market with a 400+ page book.
The lawsuit also, very notably, did not include the prompts and contexts of those prompts. Without those it would be hard to determine how much the user of the LLM actually contributed to any potential infringement directly. This lawsuit is a fishing expedition.
And in the end, this particular direction is *entirely* meaningless in the software context as for software there can be no copyright on "total concept and feel, theme, characters, plot, sequence, pace, and setting". It becomes entirely irrelevant for our discussion.