Why George RR Martins’s Generative AI Lawsuit Will Cost Authors Even If They Win
arkavian.comAnd if they lose, big companies get a signal that they can train on copyrighted data, and serve the resulting models, as long as they dont redistribute the models or reveal their "secret" training dataset. They will probably work to obfuscate the copyrighted parts of the datasets.
Meanwhile, open source models and datasets are left out to dry, as they could still be sued. Either legitimately for openly including the copyrighted datasets, or illegitimately when the defendant can't afford to take it to court.
For users, this is a lose-lose lawsuit.
> Meanwhile, open source models and datasets are left out to dry, as they could still be sued.
How come? If the artists lose the lawsuit, it's probably because AI vendors could prove fair use. From where I'm standing, it looks like it would leave everyone on fairly equal (albeit non-commercial) footing. If anything, it might give Open models an implicit advantage for being non-commercial and inherently novel.
Open datasets may still make the trainers more vulnerable, as I don't think the plaintifs even have proof OpenAI is training on full copyrighted material directly.
But even if thats not the case, small time trainers do not always have the resources to fight a takedown. This could given large entities like OpenAI an advantage over smaller trainers.
Materially, sure. OpenAI has every advantage over the smaller party.
The entire point of this article though is that banding together the smaller parties has big consequences. By forcing a judgement, either side could get an undesirable outcome. If a precedent is set, fair use or not, it would at least benefit the little guy and big guy alike to know where the law stands. From a purely legislative standpoint, I don't see how the needle moves further towards OpenAI's camp by clarifying the law.
Exactly. Everyone except for the companies with the resources to move their training to Japan loses if the authors win.
I meant that its also a loss if the authors lose, as it won't protect open source trainers/datasets!
Either way, I don't like this lawsuit one bit. They aren't arguing over the crux of the problem.
Why trust the opionions from this website? Are they legal experts in copyright?