the way it learns and reproduces text seems very different from how images can contain finite sections of existing art.
are they worried that people can get it to spit out the full book perfectly, like so many monkeys? if so, that means it's already online for free in a far more readable format