I wonder if big companies will now start paying shadow libraries like annas archive for direct access, to minimize publicity of how training data was acquired, like Nvidia supposedly did?

Few tens of thousands of dollars is a rounding error in Meta's bottom line but if this case goes anything like the Anthropic one, I would see it likely.

Of course it wouldn't prevent authors from asking LLM's for content from their books and suing Meta again but I imagine authors would be less likely to with less evidence.