Microsoft has a market cap of almost $3 trillion. I think they can afford to pay for the texts they use in their AI research.
Tell that to people haunted by BSA.
I hate the current copyright environment as much as anyone, but I do not abide double-standards, with a two-tier justice system wherein a corporation gets to freely enforce the draconian copyright regime against individuals while also getting to abuse individuals' creative works in ways much more egregious.
Out of curiosity, what would you describe a piracy manual as, if providing information as to where and how to illegally download copyrighted material is not it? What additional information would Microsoft have to have provided for it to cross the line into piracy? The only thing more illegal they could have done would be hosting the files themselves, but linking to files hosted otherwhere is still illegal, otherwise the loophole would make it child's play to ignore copyright laws. The Pirate Bay is the most famous example of attempting the legal strategy of "we don't host the copyrighted files, we just link to them", and it resulted in prison time for the founders.