It's not just books; most websites technically don't allow scraping content, but most of the content on which these models trained was scraped from the web. It'

It's not just books; most websites technically don't allow scraping content, but most of the content on which these models trained was scraped from the web. It's legality is still an open question.