Re: And on another news item...
This is a really big upcoming fight.
It is clear that under US law content created by machines is not copyrightable. And under the "moral rights" approach often advanced in Europe it should be clear that morals are irrelevant for machines. So, content created by machine should have no copyright at all (even though some people choose to pretend that there is no such thing as content without copyright).
The bigger issue is when there is some human involvement: design of algorithm, effort spent teaching algorithm, facilitation and setup of creation process, selection of outputs, editing, etc. On the other hand, some human contributions are clearly not relevant for copyright: turning the handle on the sausage machine, etc.
If we go back to the purpose of copyright: to promote the creation of content by making sure that creators are fairly compensated. We certainly want to encourage people to create interesting content-generation algorithms. But we don't really want rewards to be based on how many times you turn the handle and create a copy with a few differences (a different colour palette for the same picture, or a different genre for the same novel or movie).
What we need is some clear leadership and thinking about these new issues, otherwise it will be left to the legacy content industries to write the agenda. And their goal will be to maximize payments for whatever they think they can do (and no interest in anything else).