post
Copyright needs to be shorter for a lot of reasons, but one of the main ones would be that if copyright was a reasonable length (15 to 30 years), this would all be a moot point and everyone could use books from within a recent timeframe for any kind of use, including AI training.
Further, this would make it feel a lot less hypocritical like piracy is okay for giant corporations and their products as long as they make oodles of money but piracy for regular people is still bad. I mean for fucks sake they put the guys from The Pirate Bay in prison for less, under the same argument: that they were profiting off of pirated works, just like AI companies are. Yet somehow it's totes okay for OpenAI, Microsoft, Google, Nvidia, Meta, and Twitter, and they're not being sent to prison over it. It really re-enforces the feeling that there's two justice systems, one for the obscenely rich where as long as they make crazy profit they can do whatever the fuck they want, and one for the poor where they get fucked six ways to Sunday for doing on a small, individual level what giant corporations do at industrial scale.
The solution is to fix copyright and make more works public domain and then nobody is going to prison and nobody is getting a free pass over what's considered illegal for others.
Could you link a or the source?
The post links an image, and the description links a page with a login screen.
This is probably supposed to be a link to this: https://torrentfreak.com/nvidia-contacted-annas-archive-to-secure-access-to-millions-of-pirated-books/
There are a few communities where that link was posted correctly: https://discuss.tchncs.de/post/53133358
Sorry. I'm not sure what happened here. I thought I copied the link last night, and it seems to work for me when I click it now. Maybe a mod fixed it.
I just hope that nvidia seeds their torrents!!
Anna’s Archive is the perfect place to find specific translations of ebooks. Something I hadn’t thought of the need for until recently.
I don't know why I find it. Absolutely hilarious that Nvidia, a company currently in the AI business, notorious for not giving a shit about copyright, or just straight up going to Anna's archive.
Nah, it's pretty simple actually. If the archive doesn't exist at all, they can't even steal from it.
Would be really nice if AA during the deal was able to get docs from Nvidia like they did with music from Spotify. Like source code that could be used for drivers on Linux or older cards that aren't updated anymore. Schematics would also be fun to see. Not really for AMD, Intel, or some of the larger Chinese homegrown companies to use. But for people that do hardware repair.
Obviously drivers wouldn't be able to be offered by the distro repos or the major FOSS drivers for legal reasons. But maybe separate patches that could be applied to the less functional FOSS ones by the user. Maybe help with some of the projects that use software to make non-Cuda cards able to run programs that would require modification to run without Cuda. But I know extremely little about that outside of knowing the exist.
Since Nvidia is fine with getting pirated books, seems like they would have a fun time trying to sue for code piracy. None of the above is easy and straight forward, just would be funny to see happen.
Have you seen the quality of some of those OCR scans? I'm reading the Stainless Steel Rat books from Anna's Archive right now, and the number of errors is ridiculous, and it's not an isolated case. Pretty much every one I've read had at least a few. Good luck getting decent training data from them.
Pirating books is helping AI... this is going to cause some double standards
Incoming court decision, legal for corpos and rich fucks, illegal for plebs

1 comment