Yes, it was originally a legal dispute. Universal Music Group (UMG) and other major labels sued Udio in 2024, accusing the AI music company of using copyrighted music without permission to train its models. However, the case was not decided in court but recently (at the end of October 2025) ended with an out-of-court settlement.

 

Sadly, whether these startup companies succeed or not, the cat is out of the bag. The technology itself will never go away … unless all of the models are somehow seized and destroyed (impossible). There will always be ai music from now on.

 

It is the same battle than working class has done for centuries ... and we already know how they ended.
https://grokipedia.com/page/Luddite

At the end, AI will simply eat everything it can, like industry done with manual workers.
"Virtual internet world" was the last portion of the world without "automated" revolution.
It is going to be faster than ever in human history.

1st agrarian revolution : millenaries
2nd industrial revolution : centuries
3rd services (tertiary) revolution : decades
4th AI revolution : less than a decade

Remember : all this started around 2021 only.
Less than 5 years ago.

Amazon is kicking 14 000 ppl, UPS 34 000 ... only THIS month.
Do the math in years passed and to come.
https://www.nbcnews.com/business/bu...-ai-are-companies-actually-getting-rcna240221

ppl thinking they can "stop it" are like luddites (read above).
I can already tell you "big studios" are going to loose their fight in the long run.
And actors too. And music industry too. And ...

 

The impact of replacing the currently employed with robots or software will be in the very near future much more devastating than it was in the early 1800's when looms were mechanized. There are now billions worldwide who will be directly negatively impacted. There is no plan at all of how to cloth, feed and shelter those with no income. It's just a few years away.

 

We are so done, bro. 147k followers on Insta and a multimillion-dollar record deal:
https://www.instagram.com/xania_monet/

Xania Monet is the first AI-powered artist to debut on a Billboard airplay chart, but she likely won’t be the last
https://www.cnn.com/2025/11/01/entertainment/xania-monet-billboard-ai

 

Last edited: Nov 1, 2025 at 2:27 PM

OMG sooo good

At least they fought between them for a while. I know, not much of a consolation

 

That honestly just seems like they're trying to push the narrative
The same guys and investors whose money's deep in AI music, they promised success and they still waiting for it to blow.

They simply can't profit off AI music if it isn't popular, the system always relies on the audience.
That's probably just an attempt to make people THINK they should be listening to Xania Monet, coz she has a "deal"...
Its the old trick, fake success to get success.

Timbaland is trying really HARD to push his AI artist, with very minimal success
Even with his top connections, money and every tool at his disposal to make it work.

It would be pretty insightful to know how much % of the music industry yearly revenue is generated by an audience who does not care at all about the artist... That's where AI music could tap into

Think of those AI influencers that blew up last year on Instagram. They even got paid to be in very popular brands...
(back when AI humans looked VERY fake)
since there was no such person, people never engaged about it.
It was popular simply out of curiosity, not real connection.

Meanwhile Justin Bieber can post a picture about a fly on the wall and get 3M likes. That's where the money is. numbers do matter
Of course he's a huge influencer, but the same principle can be seen at a much smaller scale.
AI is simply not engaging for brands, luxury fragrances still invest HEAVILY into getting the latest face to promote their new product.
Rihanna did 30 times her music money with Fenty, the brand only worked thanks to her influence in music

You won't hear much in the future about "Xania Monet"... it was a flop from the start
If only the song on its own mattered, we wouldn't see the same artist have 2-3 songs simultaneously on Top 50
That's the artist itself trending, not the song

Of course i could be wrong, I'm not an expert or anything in the business

 

They're working on one in conjunction with Udio.

 

Suno's next in line.

 

This is the most important point. We all know what it can do now and eventually some open source model trained on all of the copyright stuff will emerge. It can't be stopped. Maybe 100% AI generated songs can be detected but if you create in AI, dump to DAW, and your own instruments/flavor/touch it will be undetectable. These lawsuits are making tons of money for lawyers and in the long run will be ineffective.

 

If you use these ai services and then "dump to DAW", they are still detectable not only because of the audio indicators but also the midi and the file itself has identifiable data attributes all of which you cant erase without completely ruining the files themselves so using your method you must recreate everything with your own sounds and then completely delete all ai files from the project before you render to audio or it will still be detectable. and as this thread title indicates these ai companies are now working together with the streaming and sales sites licensing their detection algos or even totally integrating with the commercial platforms to flag ai songs. Another thing is governments around the world are putting restrictions on copyrights and royalties with ai content which I think will only get worse. So is it really worth all the trouble and money to "make" a song that you can never claim as yours or ever get paid anything for it? You are becoming a slave to the major label and streaming platforms who get paid for your uploaded content regardless.

At the point of doing all that to "make" a song, it is all kind of pointless I think. You might as well go ahead and "dump to DAW" your favorite song and recreate that completely from scratch then "flip" your remake to something unrecognizable, in other words a remix, but not labelled as such and 100% yours to do with what you want and never worry about any watermarks, copyright or other file ID's these ai scumbags or major labels use, who btw also own the streaming companies. What a stupid racket!


One final thought before I go "dont feed the credits monster!" -quote by, I forget who said this in another ai thread here. Youre wasting your time and money and contributing to the ongoing destruction of professional music that isnt controlled by major label system.

 

Not trying to be argumentative, but rather genuinely curious, do you have a source for the information you listed above? As far as i have found, and i very well could be wrong, when you start to chop, transpose, eq, add reverb etc it will erode the identifiable attributes in the audio. To my knowledge, there does not exist an algorithm that can withstand such destruction. If you have any further information or reading i can do please share!! As far as midi, while there may be identifiable data in the midi file itself, that data would fail to be recognized as soon as you turn the midi into audio by running it into a synth etc. Again, not trying to be argumentative. I have done a fair amount of my own research, but easily could have missed something and would love to learn more.

Im also not trying to be a shill for AI, im rather indifferent on it for the most part. If somebody is prompting AI, then releasing that direct output as their own, then ya, that's pretty dumb and i think we can all agree they are not a producer. If somebody uses AI as a launch pad to build around and then remove, or to create sample fodder to warp, bend, and chop, i really don't care. Although of course, there is the ethical factor of the AI's being trained on works not approved for training.

I actually got suno to come within 90% of accurately recreating a song for an artist i produced. The song has moderates success, 30 million streams or so, and i wanted to see if i could determine if suno trained on something i had a hand in creating. It took just a few generations of prompting and revision. I fed the same lyrics, and used certain keywords etc and yah it was too close. Even the vocal delivered by suno had near identical characteristics.

Im hoping to see more so-called "ethical AI" come about, but we'll see.

 

After spending months on Suno and spending more money than I am comfortable admitting to here, I tried asking them directly about all of this and they wouldnt answer me. They always deflect the question and change the subject using words which really have no meaning and dont answer the question about these identifiers. I paid for the product but they wont answer simple question about their products so I know I am safe when I use it? its crazy. I then spent a week searching and using ai questions + answering and found lots of links to information about the techniques they use for identifying the content they generate and sell. These ai generation companies are very secretive about this because once there is a "fix" they can no longer make money from people for credits they sell or license their software to streaming companies and stores, it becomes valueless at that point. How else can they know if your song on Youtube or Spotify uses ai parts or the whole song is ai. How else can they detect if you have free, pro or premium account and paid for the credits to make the song or parts. It isnt difficult to understand they built these models for this purpose from the start.

The conclusion is, and yes I searched, read and asked many sources about this, there is no way to remove the audio or data based identifiers from whole songs, stems or midi without completely ruining the file to the point of it being static noise. Why th pay for ai if you can only use noise or totally different sound? we can make that ourselves for free in 5 minutes Even rerecording, resampling, chopping up, transposing, using reverbs, eqs, none of it will completely erase that identifying information they embed in the files. As I said before you can only recreate everything yourself from scratch then delete everything ai generated. This is the only way to use it safely and to ensure you dont get screwed over and in trouble with the laws around the world whether its today tomorrow or 10 years from now. Its not worth the time, money or risk I think. Its better to only use ai for an idea and then do everything yourself from nothing and delete all the ai.

I wish someone told me all this before I wasted money, so I am doing the same now for others that is all!

 

But I must ask, as you give advice to "don't feed the credits monster", etc..
Why on Earth would you even try to use it, and then spin around and give moral advice?
You complain that you can't use the ai as your own creation, and then moan about how the big ones will steal the ai...
WTF.

 

If the audio leaves the computer as analog audio, and is recaptured back into the computer again via an ADC, there is no way their "special frequencies" are invulnerable to all the processing it might be subjected to by normal production work. Nevermind someone actively out to avoid detection.
Most of these people trying to pawn off generative AI output are not proactive enough to make the music themselves, so that sort of work is usually beyond their level of effort.