> This chain has been using expropriate, not extract. Not sure why you mentioned that.

Sure, my bad.

> They stole copyrighted work without permission.

While we're nitpicking language, you can't "steal" copyrighted work. Theft/stealing are crimes that apply to situation where property is taken such that its rightful owner no longer has it. Copyright has nothing to do with theft. There is no taking of property involved. Copyright is a violation of a limited monopoly granted to a person/company to be the only one allowed to make copies.

Anyway, for a more substantive point, I think it remains to be seen that training an intelligence on information is the same thing as violating a copyright. For example, nobody would claim it's a copyright violation for a budding author to train his skills by reading tons of books.

But even setting the law aside, just morally, it doesn't seem like it should be any sort of violation. Of course, many people don't like that AI companies are getting rich, but that should be irrelevant. The fact is that nobody is harmed or deprived. Nobody's work is even redistributed. And prevention is easy -- if you don't want people/companies/AIs/etc learning from the material you put out, then keep it secret.

What seems morally bankrupt to me is this idea that anyone should be entitled to control what others are allowed to learn from. If you put information or work out into the public, that should not make you some sort of God who should control whether other people can learn from it or not.