I am not a lawyer but that is probably somewhere between the realm of derivative works [1] which the original author would have to approve and plagiarism as computerliker mentioned. I am only guessing but I would expect the court to determine how much of the original underlying work was used to create the new piece.
Might ML/GPT become a boon for copyright lawyers?
[1] - https://en.wikipedia.org/wiki/Derivative_work