The main question is why aren't they leaking it to AA themselves? Trying to keep an edge with their training sets? Isn't it ridiculous, considering the sheer size of them and statistical insignificance of the set differences?

> The main question is why aren't they leaking it to AA themselves?

How is this a question at all? They’re scanning books because the courts determined that it’s the only way to use that data. They are forbidden from using digital copies found on places like Anna’s Archive. They must acquire and scan the book.

They cannot redistribute the book. The Internet Archive tried that and the courts shut it down. You cannot scan a book and share it without violating copyright law.

> The main question is why aren't they leaking it to AA themselves?

Why on earth would they? Ultimate point for these companies is to make a ton of money, obviously they won't shoot themselves in the foot and give away whatever advantage they have, especially not to a free archive which is about doing good in the world, which probably isn't profitable enough for a company to care about.

Because that’s illegal.