They cite Kaggle as the dataset source, but that entry reads:
About Dataset
This dataset contains 7 txt files of 7 books of Harry Potter. First, I downloaded the ebooks and then converted them to txt files.
I removed the front page and the ending lines of the books to make it more clean.
This dataset can be used for NLP tasks like text processing, text mining, etc.
contact - smaindola90@gmail.com
Usability
10.00
License
CC0: Public Domain
So the dataset is public domain..? Is that how copyright works?