Recently, I’ve heard a few well known musicians declare that artificial intelligence has “stolen” their life’s work. The argument goes that because AI has been trained on their recordings, it can now churn out songs in their voice or style, hollowing out their craft and income.
I don’t buy it.
The truth is, AI hasn’t raided anyone’s catalogue; it’s absorbed everything. That’s how it functions. Just as a young saxophonist devours Coltrane solos or a bedroom producer dissects Motown grooves, an AI has to learn from the whole landscape of music, from Bach right through to Beyoncé. That process isn’t theft – it’s learning, the same way every musician has always learned: by listening.
Learning vs. theft
Where the line should be drawn is at the point of commercial use. If someone instructs an AI to “write a Tina Arena ballad in her voice” and then sells it for a toothpaste advert, that’s not fair game. But the solution isn’t to ban AI from training on music; it’s to regulate its output.
A DNA for AI music
Here’s one proposal: every piece of AI-assisted music released commercially should come with its full prompt history – a DNA trail of how it was generated. Platforms that host AI music must store this data, ready to produce it if there’s ever a dispute.
If Tina Arena hears a suspiciously familiar track on television, she or her lawyers could request that DNA record. If the prompts explicitly named her, or sneakily insinuated her style, then case closed: the work is an unauthorised copy and cannot be monetised.
Prompt data doesn’t chill creativity; it clarifies intent. It turns fuzzy arguments into facts.
Of course, music doesn’t exist in a vacuum. If the track is part of an advert that also hints visually or narratively at Tina Arena, then the lawyers may still argue misappropriation. But the crucial point is that the music itself becomes much simpler to judge. Prompt data means we can know, almost instantly, whether copying was deliberate or coincidental. In the past, such disputes involved expensive lawyers and hired experts debating “similarities” for months. Now the evidence could be hard data, not opinion. AI doesn’t weaken copyright protection – it makes it stronger and cheaper to enforce.
Cheaper, stronger copyright
And this is where the economics matter. Musicians sit at the very bottom of the industry food chain, often surviving on scraps of their own work while managers, labels, rights-holders and middlemen carve up the bulk of the profits. That lived reality breeds a kind of defensive reflex — after years of being forced into scavenger mode, every new technology feels like the next threat waiting to strip away what little remains. The anger is real, and it comes from a long history of exploitation. But it’s also why the conversation needs reframing: AI does not have to be another predator. With the right structures, it could become a lever that shifts power back toward musicians, rather than further away.
By cutting legal costs, by making copyright disputes faster and more transparent, AI doesn’t just protect artists from copycats – it saves them money. And that money, if distributed fairly, could flow back to the people who actually wrote the songs rather than the faceless executives. AI may not solve the exploitation at the heart of the industry, but it could ease one of its long-standing wounds.
Economics & power
And there’s another uncomfortable truth: some of the first work AI is likely to replace is the kind musicians often take on out of necessity rather than choice. Think of library music – those vast catalogues of background tracks for adverts, corporate videos, or soundscapes. For many, this kind of work isn’t where their creative ambitions lie; it’s a way to make ends meet in an industry that too often fails to pay artists fairly for the music they most want to create.
Library music & accountability
If AI fills that gap, is it really a tragedy? Imagine a world where musicians are fairly compensated for their albums, gigs, and publishing rights. They wouldn’t need to churn out knock-off jingles or formulaic mood pieces just to keep the bills paid. And if advertisers want a “Seven Nation Army–style” riff for a car ad, today they already ask human musicians to ape it – with no proof left behind. At least with AI prompts, there’s a record: if the advertiser really typed “make it sound like Seven Nation Army,” the evidence is there. AI doesn’t erode accountability; it increases it.
So what looks like a wolf at the door may actually be a sheep in wolf’s clothing. The work that disappears first is the work most musicians never truly wanted to do in the first place – library music written out of financial necessity rather than passion. (And if you’re bristling at that, it’s worth asking whether that feeling comes from genuine love of the work, or from being conditioned by a broken system that leaves you little choice.) In a fairer world, where artists were properly paid for albums, streams and gigs, few would mourn the loss of this grind. And the money saved in disputes could, if we fix the distribution problem, finally reach the artists instead of vanishing into the pockets of lawyers and labels.
AI as creative partner
Meanwhile, AI could open new doors in the creative process itself. The creative process in music has never been linear – it’s built on trial, error, and the courage to scrap and restart. Musicians have long relied on a process of constant iteration: testing ideas, throwing some away, and chasing the next spark. AI is not a replacement for that process but a tool to enrich it. Just as producers shape an artist’s sound – sometimes as collaborators, sometimes merely as technicians – AI can serve as an infinitely patient producer, suggesting chords, melodies, rhythms, textures. It can nudge an artist past writer’s block, or offer variations they’d never have thought of.
The final judgment still lies with the musician: to accept, reject, or reshape what AI proposes. To declare that artists must avoid these tools for fear of “contamination” is to misunderstand creativity itself. Every song ever written is built on the echoes of others.
A chance to rebalance
The row over AI and music is getting louder by the week. That’s hardly surprising: change on this scale always breeds suspicion, fear and grandstanding. But too much of the debate is being driven by anxiety rather than clear thinking. A world with AI doesn’t automatically spell the theft of human artistry. In fact, with the right rules in place, it could deliver the opposite: faster justice in copyright disputes, stronger protection for originality, fresh space for creativity to flourish – and, perhaps most importantly, the removal of the drudgery that has long drained musicians both creatively and financially. AI has the potential not to hollow out the role of the artist but to strip away the bureaucracy, the endless legal wrangling, and the parasitic middlemen who have thrived on keeping music’s economics deliberately murky.
Bottom line
That is the real moment of clarity. The sooner artists step back from the panic and look at the bigger picture, the sooner they’ll see that AI is not the enemy at the gates but a tool to help rebalance a system that has failed them for decades. It can pierce the legal fog, expose bad actors in an instant, and defend musicians’ rights more effectively than any courtroom drama ever could. It can give them back their time, their focus, and – if the industry finally lets it – the fair rewards they’ve been denied. Instead of treating AI as a threat, artists could seize it as a once-in-a-generation chance to reshape their world.
AI won’t kill music; it might finally free musicians from the industry that has exploited them for decades.