Why Human Creativity Must Become More Valuable, Not Less

The music industry is growing. Global recorded music revenues reached $31.7 billion in 2025, up 6.4%, with 837 million paid streaming subscriptions worldwide, according to IFPI. The business is healthier than it has been in two decades. The question is who captures that value, and what happens when the source of it becomes easier to imitate.

The answer is arriving faster than the industry's ability to respond. Deezer receives roughly 75,000 AI-generated tracks per day, about 44% of its daily uploads, up from 10,000 a day just over a year earlier.

Yet fully AI-generated music accounts for only 1% to 3% of streams, and 85% of those streams were detected as fraudulent. Apple Music confirmed a similar pattern.

More than a third of its monthly uploads are 100% AI-generated, but those tracks represent less than 0.5% of actual listening. The flood is real. The audience is not.

That gap is where the argument starts. Reuters reported in April 2026 that Anthropic asked a California federal court to rule that its use of copyrighted song lyrics to train its AI chatbot was fair use. Music publishers argued that the training was not fair use because the model generates derivatives that compete with the market for their lyrics. The case is one example of a question the industry has not yet answered. Who controls the creative work that AI systems learn from, and who benefits when that work is reproduced at scale?

Marko Anastasov works on that question every day. He is the owner and CEO of Muze, a Bulgarian music company working in licensing, distribution, rights, and production. He describes his perspective as that of a business-focused commentator on music rights and intellectual property, not simply a regional executive. His argument is not that AI should be stopped. It is that the economics have to change before the industry loses the source of its value.

"Music is the most consumed culture at the end of the day," Anastasov says. "You don't need to watch Netflix for three hours. You can listen to a track for three minutes, and it can make your day." That ease of consumption is what makes music valuable. It is also what makes it vulnerable, because when supply is abundant and imitation is cheap, the price the market will pay falls. He adds, "As long as the supply is so huge, the people are going to be willing to pay less and less."

The platforms are not the primary problem. A major music streaming platform can pay out more than 60% of its revenue to rights holders, and that is a defensible rate. The problem is what happens before and after the payout. Anastasov estimates that around 47% of value is lost before it reaches artists and authors, through collection failures, rights fragmentation, and intermediaries.

His proposed split is specific. AI companies should pay when they train on copyrighted music. Major labels and distributors are already signing deals with AI music platforms, and that trend will continue. But he draws a line. AI-generated music should not appear on streaming platforms, and it should not carry intellectual property rights. AI platforms can let users create whatever they want, and a share of what those users pay should flow to the creators whose work trained the system.

"AI is about natural selection," Anastasov says. "The greatest songwriters and artists can never be replaced with AI. However, all the others that weren't good enough will be replaced by AI." That is a hard view, and it is not a comforting one. But it points to a principle the industry can act on. AI should raise the floor of what is possible for creators, not lower the value of what is created.

The risk, if the industry gets this wrong, is not that music disappears. It is that the value migrates to the companies that own the models, and the people who make the music become a smaller and smaller part of the economy. "It could easily collapse in two days," Anastasov says. That is his warning, not a forecast. But the underlying point is defensible. The value of human creativity underpins everything else in the business, and if that value is not protected, the industry is trading its own foundation for short-term efficiency.

The practical path is unglamorous. It requires transparent licensing, rights holders who share in the value AI generates, and AI systems designed to facilitate human creation rather than replace it. None of those changes are dramatic, but all of them require the industry to act before the market decides for it. If music is the most consumed culture in the world, why is the person who makes it the last to be paid?