There's no real secret. They scraped the collected works of humanity with no regard for intellectual property rights and are now hiding behind the guise of being a "non-profit" company, despite raking in loads.
I get the last part of your comment, where they make money from their AI's, but I'm curious if you could explain why them scraping everything on the internet is a bad thing? I'm really curious, as I see it more as an advantage that they could scrape just everything rather than a disadvantage: if we start getting LLMs that only scrape certain sites or certain topics, we're just making the echo chambers even more defined.
Them scraping everything feels a bit like the internet era where APIs were open and semi-unlimited.
I do get that them having used , for example, all the Tolkien books as part of their dataset, and users being able to ask their tools to "write a new LOTR" book, could be seen as a problem, but even then it's only a problem when they commercialise it, and we have laws in place for those things. Fanfics have always been a thing and are just the same, except that it takes more effort as it's a human writing it.
Again: I'm really curious about your view and opinion, as you can see that mine is quite different.