Meta shifted blame to rogue employees watching adult content on company time, but court flatly rejected the excuse
All AI models should be forced to publish everything they used to train their models. Not only should they publish it but it must be verifiable.
And fined heavily for everything they didn’t ask for first.
Instantly kills any kind of open source and let’s Disney and YouTube dictate who gets to build AI and who doesn’t.
The data isn’t owned by the little people, it’s all big copyright against big AI.
There’s fair use and business use.
A business should not be allowed to profit from the work it doesn’t own.
The laws you ask for are only going to give the business aspect (openai, anthropic, google, etc) an instant monopoly. The money won’t even go the the artists and content creators but to data brokers like YouTube and Reddit. The whole anti-ai movement is a joke.
TLDR: Meta now faces lawsuit, alleged to have systematically torrented copyrighted porn, likely for Ai training. Meta’s defense arguing the torrenting were done by rogue employees having a wank was not accepted by the judge because of keyword<->IP range patterns suggesting systematic scraping.
IMO, the funny / sad thing is that it’s the torrenting that got them. Not the, IDK, scraping of every human work ever put onto the internet for the explicit act of producing derivatives, even in cases where the original author has indicated that they don’t want that thing. No, no, it’s the torrenting that’s devil’s work.
If they are training their AI on porn videos, then they are expecting people to use their AI to create porn.
The next step is finding out what kind of porn they are allowing.
They are able to not only detect if you have a hard-on, but also the exact location and orientation in your pants and your actual penis size.
