A few months ago, I came across a video on YouTube where someone had taken 1969s Led Zeppelin II album and re-recorded it into what the album would have sounded like if it had been released in the 1950s. As soon as Whole Lotta Love started playing, I couldn’t believe my ears and like the naïve person I can be sometimes, I thought to myself, wow these people who re-arranged the album to sound like this are amazing. That takes a lot of ingenuity and talent. And then a few days later it dawned on me that it was done using AI and the joke was over, actually to be more precise, the joke was on me.
I’m not always on top of everything that’s going on, especially in the AI market. I know what ChatGTP can do. Ask it anything and it will give you an answer, like some wise oracle out of a fantasy book, but I didn’t realize the technology had come this far. I know it could probably write this article or just about anything you like, but we’re way beyond that now folks.
Fast forward a few months and I stumble on a video reviewing a tool called Suno. This is a tool, along with its competitor Udio, that allows anyone to write a song. Anyone. Whether they can play an instrument, write lyrics or know anything about songwriting or music theory, for that matter.
How did these companies do it? Well, this is where the deceptive part comes in, in my opinion. I don’t think someone sat there and force fed albums they purchased into the AI program one by one, like spoon feeding a baby. No, they likely data mined entire genres of music from music streaming platforms such as YouTube and Spotify. And this is where I have an issue and where the big three record labels have an issue as well.
Music labels sue AI companies Suno, Udio for US copyright infringement.
As a songwriter for the past thirty-five years, I have to say that I am sickened by what I have seen and heard. I watched a couple of videos on using these music-generating AI tools and, like hearing the 1950s version of Led Zeppelin II, I sat dumbfounded as songs were created by simply typing in something at the prompt. So I tried it out and typed in “blues song about a bullfrog that missed the train back to the swamp” and boom, it spit out two songs. Here’s what it created as sad as it is. Not because the bullfrog missed his train back to the swamp but for the state of music in 2025.
Of course, you can also feed the beast with your own lyrics and riffs and it will write songs for you based on those ideas. That’s really great and awesome, but here’s the problem. In order for the AI to do this, they had to train the system with entire catalogs of copyrighted music, created by slews of artists, across multiple genres, spanning decades, to be able to mimic any style of music the user wants. Blues, rock, country, disco, metal. You name it, it can do it. This is where the lawsuit come in.
I watched another video by a songwriter who fed his lyrics into the beast system and it generated a bunch of songs based on the style he was looking for. It was interesting to hear a songwriter’s perspective on it. He was so floored by what it spit out that he was actually touched by how beautiful the song was. He claims he could not have come up with the idea himself. That sounds about right, because the system used all the greatest minds in modern music to draw from.
Picture this, you have John Lennon, Paul McCartney, George Harrison, Pete Townshend, Keith Richards, Ray Davies, Freddie Mercury, Elton John, Roger Waters, Tom Petty, Don Henley, Prince and Bryan Adams all sitting together in a room and you hand them some lyrics you wrote. You ask them to collaborate and write a bunch of songs in a particular style and you want them all fully arranged and recorded with a full band ensemble by the end of the day. As crazy as that sounds, that’s essentially what these tools are doing, except they are doing it in a matter of seconds and pilfering thousands of the greatest songwriters and their ideas. Melding all those minds into multiple algorithms. I don’t know about you, but I don’t think it’s cool.
Here’s the deal with songwriting. All copyrighted music is covered under what is called the Musical Work copyright (lyrics, music, melody). Of these three elements that make up a song, the melody is the most important part of a song because it has to be inspired, and more important, in light of this topic, it has to be inspired by a living, breathing, human being, with blood coursing through their veins who have something we like to call a soul.
It shouldn’t have to be said that there is a clear distinction between human inspiration and AI. The AI algorithm does not have inspiration. It’s nothing more than a bunch of zeros and ones residing on a hard drive array located in a cold, lifeless data center somewhere. It has no soul.
So going back to the songwriter who was touched by the song the AI generated from his lyrics, well, he used the tool to write the melody for him, is what it essentially boils down to, all feelings set aside. The melody is the hardest part of a song to write. Just like the hardest part of becoming a lawyer is passing the bar exam, well, he basically used AI to write the bar exam for him.
I could see using these tools to create the backing tracks for a song idea or helping out with the lyrics, but not the melody. Once you hear it, you can’t unhear it. This is what drives songwriters crazy because when they have a great melody for a song, they have to capture it right away before they lose it. If a songwriter is using AI to write their melodies, they can no longer really be considered the songwriter, no matter how much they want to take credit for typing an idea into a prompt.
The final catch in all of this is, anything you create using these tools, like all the copyright music that was fed into it, can be used at the company’s discretion to create other songs based on your ideas. Only seems fair, considering they probably lifted the entire Beatles catalog. If you use the free version of the tool, you have no commercial rights to the song it generates. If you use the paid version, you own the rights to the song, at least in the case of Suno. But unless you know copyright law, and I’m no expert, when it comes to music-generating AI, you can not claim a copyright on it unless you add your own creation to it. So there has to be some kind of human element embedded in the song.
This is a slippery slope we’re going down, and it opens a whole can of worms on so many issues. The most important being that the human element in creating the idea is gone. The same can be said for books and art generated by AI. Humans no longer required. A phrase I’m sure most of us do not want to get used to hearing, but there it is.
Maybe most people don’t care, as long as the song is good, but it’s the long-term effects that are reasons to care. Once you start taking the human element out of this world, what will we have left?
What’s your take on all of this? Let me know in the comments.
If you’re actually interested in learning how to write songs, please check out my kindle book on the topic, Songwriter Handbook.