- cross-posted to:
- technology@lemmit.online
- cross-posted to:
- technology@lemmit.online
Did that AI system use Doja Cat records for training data?
You must log in or register to comment.
This creative commons thing popping up to combat AI has about the same energy as those Facebook posts stating “I do not give my permission…”
Words can be copyrighted and licensed. Otherwise books and scientific journals protection are also meaningless. What some commenters do is just stating that their content is licensed under a certain term in the addition of whatever else the EULA/ToS for the service imposed on, provided there is no conflict.
Edit: Just to be clear, I am just stating the status quo and not defending it. I think copyright law needs to be reformed altogether.
With text, it’s possible to trick the AI to reveal its license or for it to just print the license unprompted (like Github’s CoPilot), but with music, how is anybody going to detect it?
What’s the point of posting that “anti commercial ai license” on every comment? It’d be so easy to filter it out of a training dataset. Or am I missing something?
It’s dead easy. Yet github didn’t do it when training copilot and are now sued because of it.
It is also easy to build a database of copyrighted material and check that revealed training data marches it. The copyright licence doesn’t necessarily need to be attached. It just makes it easier to spot.
Also, what are you arguing here? That because copyright is easy to ignore, it should be or that it’s pointless? Is that the advice you’d give anybody else too? “You know what Disney, everyone ignores copyright, so why not make everything public domain?”
Maybe musicians might end every song with some magic words that stop people using it.
This feels like one of those Ford Pinto moments, where the accountants and lawyers got together and determined there was a larger profit margin to take the path less traveled
