I wrote up all the things I do to try and slow the bot flood when some creeps decide to do another training run.
Remember: The don't have a copy of their training data. They use the public web as their training data. Their data sets are simply lists of URLs. Every AI fucker downloads them all over again in order to train the next generation. It'd be prohibitively expensive for them to keep a copy, of course, and making a copy would invoke copyright law. No good! Imagine that cartoon where the dog has a ball and thinks "THROW!" and when the human hands tries to take the ball, the dog doesn't let go and thinks "NO TAKE! JUST THROW!" This is the AI trainers. The want to argue: JUST TRAIN! NO COPY!
I sometimes wonder how the pendulum will swing regarding copyright. AI bros will argue that ingesting all the works and regurgitating it as AI slop removes all the copyright. A lot of creative works and software will be made using AI slop and so the value of copyright protection will decrease. Perhaps the copyright maximalists will lose. Mickei Muuse might be close enough to Mickey Mouse for many people.
The reaction will be to hide the real new material, of course. More of that "private screening, no cameras". More valuable software not available from public repositories. More DRM that cannot be broken by AI ingestion.
Imagine the dog in the burning building saying "this is fine." This is not fine. And the world is burning. War, climate crisis, fascism. The polycrisis is upon us and this is just yet another fire that we started.
@alex The truth is that the copyright law never really protected us. Just like law doesn't protect you on the street from a physical assault. Only if you make a lot of fuss about it, complain loudly in the media, maybe hire lawyers, there is going to be some chance of retribution. But it will be already too late for you, of course, because you would have already have been beaten. Large organizations that specialize in it, like security agencies, police, army, etc. use those laws, but not us.
@alex The only function of copyright law is to create new territory for the land owners to conquer, because they already own all the physical land. And just like trespassing through someone's private garden is not going to result in legal action, so it won't with copyright infractions for the private authors. But try to trespass on a factory, sports stadium or airport, and the situation will immediately escalate. Even though the laws, just like the copyright laws, are theoretically the same.