I'm not a lawyer, but neither is this Mike guy. I'm quite suspicious about how confident he is in stating that all of this is legally fine. From the creator him

I'm not a lawyer, but neither is this Mike guy. I'm quite suspicious about how confident he is in stating that all of this is legally fine. From the creator himself: "When I ran out of books on my own shelves, I looked to the internet for more text that I could analyze, and I used web crawlers to find more books." I'm annoyed by the phrasing. Because you're obscuring that you pirated commercial books. A commercial book is not open or public data. You're supposed to pay for it to get access to it. This assumption that just because something can be found on the internet, it's cool to just take it and do whatever the hell you want with it, is worthy of push-back. Especially in the case of data that isn't public at all (not by intent anyway). This is not the same thing as scraping Wikipedia, which allows for it and has supportive licensing for it. I believe that in the majority of cases where we're talking about entirely new use cases such as big data, deep analysis, AI training, etc we need to move to an opt-in model. Unless explicitly specified, you have no permission. The opposite situation is ridiculous. I do agree that the mob-like type of criticism deserves criticism in itself. Trigger words like "AI", "crypto", "white techbro" converting adults into sadistic bullies is a sad thing to watch.