There is an old rule in the music business. Whenever somebody starts talking about innovation, follow the money.

Last week, Spotify and Universal Music Group announced a new licensing agreement that will allow Spotify Premium subscribers to create AI-generated covers and remixes using songs from participating Universal artists and songwriters. The announcement was wrapped in the language we have come to expect from both the tech and music industries: consent, credit, compensation, creativity, fandom and opportunity.

All of those things sound wonderful.

The problem is that none of them answer the questions artists, songwriters and managers should be asking.

Who gets to say yes?

Who gets to say no?

Who gets paid?

And perhaps most importantly, who is actually in control once their music enters the machine?

For the past three years the music industry has largely positioned artificial intelligence as an external threat. Record labels launched legal actions. Artists signed open letters. Industry organisations warned that AI companies were scraping and training models on copyrighted music without permission. The battle lines appeared simple. On one side stood creators. On the other stood technology companies racing ahead of regulation.

This deal changes the conversation entirely as Spotify and Universal are no longer talking about how to stop AI. They are talking about how to commercialise it.

What we are witnessing is not the arrival of AI in music. AI has already arrived. It has been flooding streaming platforms, social media feeds and video-sharing sites for years. What makes this deal significant is that two of the most powerful companies in music have decided that the future is not about resisting AI but building a business model around it.

From a commercial perspective, the logic is obvious.

Spotify has spent years searching for new revenue streams beyond traditional subscriptions. The company has been increasingly vocal about the value of so-called “superfans”, listeners willing to pay extra for exclusive experiences and deeper engagement with artists. AI-generated remixes and covers fit neatly into that strategy. Instead of simply listening to music, fans become participants. Creation itself becomes a product.

Universal’s incentives are equally clear.

If AI-generated music is going to happen regardless, and there is little evidence to suggest it can be stopped, then the smartest move is to ensure it happens inside a licensed ecosystem rather than outside one. Better to collect a share of the revenue than spend the next decade fighting an unwinnable war against technology.

When you look at it that way, the agreement makes perfect business sense. Whether it makes sense for artists is another matter entirely.

The most important word in the entire announcement is not AI either; it is participation.

The deal repeatedly refers to “participating artists and songwriters”, suggesting that creators will have some degree of choice over whether their work is included. On the surface, that sounds reassuring. Few people would argue against artists having the right to decide how their music is used.

Yet participation is only the beginning of the conversation.

What exactly are artists agreeing to?

Can a songwriter approve a remix while refusing permission for lyrical changes?

Can an artist allow their catalogue to be reinterpreted without allowing their voice to be replicated?

Can permissions be withdrawn if an artist becomes uncomfortable with how the technology evolves?

The music business has a long history of presenting broad consent as meaningful control. The two are not always the same thing.

A signature on a contract is not the same as ongoing approval.

Artists know this better than anyone.

Many of the contracts signed in previous decades never anticipated streaming, social media, short-form video or artificial intelligence. Yet those same agreements continue to shape how music is exploited today. That is why the details matter far more than the press release.

The second issue is quality.

The technology sector has always operated under the assumption that more content is inherently better. The music business should know better by now.

Streaming has given listeners access to an almost unimaginable volume of music. Millions of tracks are uploaded every month. Discovery has become one of the industry’s defining challenges. Artists are not competing solely against each other. They are competing against a constantly expanding sea of content.

AI risks accelerating that problem dramatically.

One song could theoretically generate thousands of remixes, reinterpretations, genre shifts and derivative works. Some may be brilliant. Many will not. The question is not whether audiences can create more music. The question is whether more music automatically creates more value.

There is a tendency within technology circles to confuse quantity with creativity.

They are not the same thing.

Music has never been valuable because it is abundant. Music is valuable because it is meaningful. The emotional connection listeners feel towards an artist does not come from infinite variations of a song. It comes from the belief that somebody, somewhere, had something worth saying.

That is where the current AI debate often loses its footing. Technology companies tend to discuss music as content. Artists understand it as culture. Those perspectives are not always compatible. And then there is the money issue.

Every major announcement about AI in music eventually arrives at the same destination. Compensation.

Spotify and Universal have both spoken about creating new revenue opportunities for artists and songwriters. Nobody would object to that. The problem is that we have not been told what those opportunities actually look like.

Will payments be based on streams, creations, downloads or engagement?

How will attribution work?

How will rights holders audit usage?

What percentage of the revenue generated by these tools ultimately reaches the creators whose work made them possible?

The music industry has heard promises about future income before.

Artists were told that streaming would create unprecedented opportunities. For a small number it did. For many others it delivered fractions of a penny and an endless fight for visibility.

That history should encourage caution rather than blind optimism.

None of this means the Spotify and Universal agreement is inherently bad. In many respects it may be preferable to the alternative. A licensed framework is certainly better than a lawless one. Artists having some degree of control is better than having none. Compensation, however imperfect, is better than uncompensated exploitation. But we should resist the temptation to celebrate before the details are clear.

The real significance of this deal lies in what it reveals about the industry’s direction of travel. The battle is no longer about whether AI belongs in music. That battle has already been lost.

The new battle is about governance. It is about ownership. It is about consent. It is about whether artists remain active participants in the future of their work or become raw material for systems designed and controlled by others.

For years the music business fought to establish that creative work has value. Now it faces a more complicated challenge: ensuring that value remains attached to the people who create it.

Because if the future of music is going to be built on artificial intelligence, artists deserve more than vague promises about innovation.

They deserve clarity, control and a fair share of the rewards. Anything less is simply the latest version of an old industry story.

Words: Linda Coogan Byrne