Understanding AI Music Generators Copyright Argument (Udio and Suno) + Moshi AI 

When people say AI will take over the world… “I’m not sure” 😂😂😂😂😂😂😂😂😂😂

So hilarious 🤣🤣😅 1:30

The “evidence” in the RIAA lawsuit doesn’t prove anything because the testers input the lyrics of those songs using the custom lyrics feature. The platforms didn’t spit out the lyrics automatically. Also, all music is based on patterns (chord orogressions, beats, melodies, etc. ) The training was on millions of tracks, therefore it learned pretty much everything. Input the lyrics then just keep generating until you get the right patterns for each song section. You have the option to run each section separately. This isn’t hard to do. They didn’t generate one time and there it was. They could have generated each song section hundreds of times. Chain out those sections over time and there you have it. I heard these examples and they are NOT exactly the same as the originals. And if the claimed examples turn out to not be in the training data, that’s the end if that. AI doesn’t sample the music and doesn’t need to store it. Storage wouldn’t be a problem anyway. The RIAA just needed to present some probable cause to get the case into court so they can get the platforms to reveal their sources, hoping that some of their catalogs are in there even if none of the examples are there. The platforms seemingly admitted to using at least some copyrighted material because they’re claiming fair use. This type of traing as never been addressed by fair use and even the US Copyright office has not made policy in traing use. They could possibly win in that grounds because the use can be covered by fair use. It all depends on how it can be explained and who hears the argument. It is NOT open and shut like the RIAA wants us to believe in their court documents. My guess is that the RIAA will offer to settle for a revenue sharing deal before this gets to court and take an on-going revenue share. That is probably the goal. The labels have contracted other AI companies to do exactly the same thing as these AI companies with the sample music being used for training. They are trying to eliminate their competition or take part of their revenue. There is no real concern over damage done by AI or they wouldn’t be doing it themselves. This is all about greed. As for the examples being cooyrighted infringement, that is on the people who release the outlut that they oromoted, not the AL companies. Guess who did that. It wasn’t the AI companies. It was the testers who did it on behalf of the RIAA. They promoted the infringing works and then released them.

6:50 is what you meant to say here “right of publicity”? Styles can’t be copyrighted, and likeness is a separate area of law. Could you ask your legal expert to consider what happens if consumers collectively say, “nah” to the music industry’s offerings in this area? That’s what effectively happened in the 00’s, and given that there are/will be AI sound generators runnable locally by end users, may render any licensing paradigm unenforceable.