You're never going to get a textbook out of an LLM, or a chapter, or a page. Big-IP lawyers probably spend weeks engineering prompts to find enough stuff that comes out un-paraphrased. Poems and lyrics, maybe.
The legal test for fair use considers "the amount and substantiality of the portion used in relation to the copyrighted work as a whole." In this case it's the portion regurgitated near-verbatim, since paraphrasing and summarizing is fair use. You can write Cliff Notes without the author's permission.
LLMs are probably capable of failing the fair use tests for less than 0.1% of the stuff they're exposed to.
The legal test for fair use considers "the amount and substantiality of the portion used in relation to the copyrighted work as a whole." In this case it's the portion regurgitated near-verbatim, since paraphrasing and summarizing is fair use. You can write Cliff Notes without the author's permission.
LLMs are probably capable of failing the fair use tests for less than 0.1% of the stuff they're exposed to.