- Disney and NBCUniversal have teamed up to sue Midjourney.
- The companies allege that the platform used its copyright protected material to train its model and that users can generate content that infringes on Disney and Universal’s copyrighted material.
- The scathing lawsuit requests that Midjourney be made to pay up for the damage it has caused the two companies.
Piracy to me is not the same thing, I’m actually not in favor of piracy, because the way I see it if you want access to a content, and the creator says that you need to pay for that content then you will pay for it. If not then you don’t really want access to it, or you in fact simply did not want to pay for it in which case it’s very similar to stealing. None of the pro piracy arguments convince me, except the ones in which it’s about consuming the content in the format that you want. Ie I buy books from Amazon, but only because I want the writer to get their cut, but I will either remove the DRM off the book or pirate it. So yes if the AI was trained using for example a book whose content is not freely available and ChatGPT simply pirates the content of the book to train their models, then they are in the wrong.
But here’s the thing about my argument regarding AI training data. I never played Warcraft3 nor World of Warcraft! I only saw the cool art that was displayed on GameStop, online and on shirts on hot topic. I never paid Blizzard for access to Arthas, the design of Arthas was publicly accessible to me by virtue of Blizzard trying to promote their game. So I guess what I’m saying is if the content they trained a model is publicly accessible to people without payment, then there’s no reason AI cannot be trained on it.
As far as the ethics of piracy, my stance is that The current model where we are essentially paying for someone to hit copy and paste is inherently broken and we need to move to a model based on commissions/grants etc where the artists are being paid to make the works, ideally through public funds.
As far as the current reality around the AI companies, that Arthas stand was one gamestop had a license to display in their store provided by Blizzard to promote their game. The copyrighted material you “trained on” was provided for piblic access. These companies aren’t training off of publicly available information, as I was saying, they’re torrenting the full copyrighted material in the exact same way a traditional pirate would, The only difference is what they’re doing with it afterwards.
Edit: I don’t know if there’s evidence of Midjourney’s developers doing it specifically but Meta absolutely has so it’s a pretty safe assumption.
Terrible idea man. Can you imagine Trump being in charge of funding all of the arts? I don’t want any government with that sort of power over creative endeavor.
The current system works. You’re not paying for someone to hit copy paste, you’re paying for access to the idea that is physically embodied in the content if that makes sense. The creator decides whether you pay for that or not, and how much to pay. But many pirates don’t want to pay, don’t want to watch ad; in summary they simply believe that they are entitled to the work of the creative, which to me is absurd and outrageous.
But yeah that’s what I meant about AI training. If there are Shrek images out there that Disney willingly published and I trained an AI on it there should be no issue because it would be no different than me looking at Shrek and then making a drawing of it.
Yes but looking at publicly available shrek images is not what’s happening here, this is downloading every shrek movie
Yeah I guess that’s what they’ll have to prove to win this.