If they were sharing the scans it would be great but they’re not.
Google Books tried. They lost. If you want this fixed, talk to your congressman, not Amazon.
And if they're not out of copyright why can they train on them without permission?
They probably argue that training is fair use, but if I take a film and compress it with a novel data compression codec and distribute it I'm violating copyright. LLMs are lossy compression for textual information.