People would have at least somewhat less of a problem with this if they also put up an archive of PDFs of all these rare books if they are out of copyright.

But that would help competitors with training data, which I assume is why they don’t do this.