MBW Views is a series of op-eds from eminent music industry people… with something to say. The following was originally delivered as a speech at the United Nations’ AI for Good Global Summit in Geneva on Wednesday (July 8) by ABBA co-founder and CISAC President Björn Ulvaeus.
Turning the summit’s name into a challenge — “AI for Good… good for whom?” — Ulvaeus offered a test: a technology is only good, he argued, when the people whose work made it possible aren’t erased by it, “when they consent to it, when they share in what it creates.” He said he already lives inside an example of AI done right — ABBA Voyage, the London residency where digital avatars of the group perform six shows a week, made with the band’s consent. “It isn’t a slogan; it is a contract.”
Generative AI, he argued, has no such contract: music models have been trained on “a century of human songs,” his own included, “ingested without agreed permission, without payment, and without so much as a postcard.” But Ulvaeus, who uses AI almost daily to build demos for a musical he’s writing, said the answer isn’t to trace creators’ work in the outputs these systems produce, but to pay for the inputs.
“We should be paid for what went in,” he said, “for the raw material that made the machine what it is.” His proposed fix borrows from streaming: a share of AI subscription revenues flowing back to creators collectively. What’s missing, he told the room, isn’t the infrastructure but the political will “to require it for everyone, not just those powerful enough to sue.”
Here is Björn Ulvaeus’ speech in full…
Good morning.
Last month, I stood on a stage in Paris, marking 100 years of CISAC, the Global Confederation of authors and composers, and this month I stand on a stage here in Geneva, and I find that the journey from one city to the other has a story to tell, because Paris is where the rights of creators were born.
In 1791, when France became the first country in the world to write into law that the work of an author belongs to an author. And Geneva is where those rights became universal. The treaty that says a song written in Stockholm is protected in Seoul, in Sao Paulo, and in New York. The Berne Convention has its guardian here at WIPO, just a few kilometers from this stage.
Paris made the promise, Geneva made it universal, and I’m here to ask this summit to do the same thing one more time, for the age of AI. The name of this summit is AI for Good. It’s an inspiring name. This week, the United Nations launched a new AI for Good Commission, bringing government scientists and technology companies together to help shape the future of AI. I hope one extra chair will always be reserved at that table for the creators whose work made these systems possible, because the title of this summit contains a question: Good for whom?
I want to suggest a simple test. A technology is good when the human beings whose work made it possible are not erased by it, when they consent to it, when they share in what it creates, by that test AI for good already exists. I know because I’m standing inside an example of it every night in London. Some of you may have seen ABBA Voyage, four digital avatars, myself among them, 40 years younger. I can recommend that. Performing usually to a full house, six shows a week, made possible by motion capture and machine learning.
Some people sometimes ask me, “How can you lecture the world about AI and human creativity, and then sell tickets to watch a machine perform as you? The answer is one word: consent. We chose it. We participated in it. We are paid for it. The technology serves the artist because the artists were at the table from the very beginning, and the audiences love it.
So, I guess that’s AI for good. It isn’t a slogan; it is a contract. But there’s another version, one without the contract. Somewhere in the world today, a generative model is producing music. It produces more songs in a day than every member of CISAC will write in a lifetime. It can do this for one reason only. It was trained on our work, on a century of human songs, including, I have no doubt, mine. Ingested without agreed permission, without payment, and without so much as a postcard.
I’m not going to stand here and tell you that machine-made music is cold or soulless, and that audience will always hear the difference. I know that isn’t true. I am genuinely in awe of the tools that have been built, but awe is not the same as acceptance. The tools are extraordinary, but they could not have been built without us – the artists, the songwriters, the musicians, the producers, every person who has written a melody, recorded a performance, found the words was something that perhaps had never been said before. Their work lives inside these models.
“I have lived through the cassette, CD, the MP3, Napster, streaming. Every time our industry found a way to embrace new technology while protecting the human creator.”
Björn Ulvaeus
It is the foundation on which these tools were built. It is the reason they sound like they sound, move the way they move, and feel the way they feel. That makes us partners, not protesters, not obstacles, not a problem to be managed by lawyers. Partners. And partners deserve a place at the table. Partners deserve a share of the harvest.
Now, what does that actually mean in practice? Because I’m not here only to make moral argument. I’m here to propose one. Some people argue that the answer lies in tracing our work in the outputs, and I understand that instinct. But I think it misunderstands how these models work.
As we all know, they don’t sample recordings; they learn relationships across billions of examples, how melodies move, how harmony resolves, how rhythm, timbre, and language combine. What comes out isn’t a copy of any one song. It is a new synthesis built from everything the model has learned.
Now, let me give you an example of how I work with AI. I’m writing a new musical at the moment, and I’m using it almost every day. I record myself singing and playing a song or a part of one, which I’ve written; I’m playing it on my guitar. I’ll upload it to the AI model, and I ask it to give me a cover because I want to hear the song roughly as it’s going to sound when I someday hopefully record it with a human artist, so I simply prompt, I ask it to give me pop ballad, female vocal, strings, and suddenly I have a full-blown demo.
Weaknesses I couldn’t hear before suddenly become obvious. A bridge that doesn’t quite do the job, words that don’t sound right, a melody that needs another turn. Then I rewrite, and I do it again and again. I challenge anyone to trace that back. You can’t.
For me, tracing the output was always the wrong question. The right question is much simpler. It is about the training. Our works went in. We should be paid for what went in, not for every output that comes out the other end, but for the raw material that made the machine what it is.
We’ve solved problems like this before. When Spotify emerged, we didn’t try to measure the value of every individual listen before paying creators. We licensed the catalog. A percentage of the platform’s revenue flowed back collectively to rights holders. AI can work the same way. A share of AI subscription revenues could flow back to the creators whose work trained these systems.
Managed collectively, just as collective licensing has worked for more than a century. The infrastructure already exists. The principle is already established. What is missing is the political will to require it for everyone, not just those powerful enough to sue. And that is what I’m asking this room for today.
This is not simply about protecting today’s artists. It is about making sure there are artists tomorrow. If human creators cannot earn a living, fewer people will devote years to mastering an instrument, finding their voice, or writing songs that matter, and if that happens, AI itself will eventually have less original news, human creativity from which to learn. Fairness is just not morally right. It is how we keep the well from running dry.
We are sitting right now at a fork in the road. The legal cases now moving through courts around the world will help determine whether licensing becomes the foundation of this new industry, or whether free extraction does. The decisions made over the next few years will shape the creative economy for decades and decades. I’m not pessimistic. I have lived through the cassette, CD, the MP3, Napster, streaming.
Every time our industry found a way to embrace new technology while protecting the human creator. Every time, people had to make difficult decisions. And every time the decisions that lasted were reasonable ones. Human creativity is not just content; it is not just data.
It is testimony, a life lived, a grief that became a lyric, a love that became a melody, a specific, unrepeatable experience of a human being in a particular time and place, transformed into something another human can feel that is what trained these models, and the people whose lives work became part of those models deserve to share in what they have made possible.
So I say to the companies building these extraordinary systems: you have built remarkable things. You could not have built them without us. That makes us partners. We deserve a place at the table. We deserve a share of the harvest. Human creativity is not the enemy of artificial intelligence. It is the reason artificial intelligence exists.
The future does not require us to choose between creators and technology. It requires us to decide whether creators remain partners in the future they help build.
Paris made the promise. Geneva made it universal.
The people in this room have the opportunity to renew that promise for the age of AI.
I believe we will.
Thank you.Music Business Worldwide
