Whistle of the Signal

EN SV

Whose Hand Is on the Instrument?

AI, Music, and the Question of Authorship in 2026

Something shifted in the last year. AI music tools stopped being the party trick — type a mood, get a jingle — and started behaving like something closer to a bandmate. Models now read for emotion and narrative, not just notes and tempo. Voice conversion can carry a singer’s phrasing and breath into a completely different tone. And by mid-2026 the licensing side finally caught up with the technology: Warner Music struck the first landmark settlement-and-license deal with Suno, and BMG followed in August with a global publishing-and-recordings agreement built entirely on artist opt-in — BMG’s EVP called it a deal with “clear economics and new creative possibilities” for the artists who choose in. Universal, notably, still hasn’t signed with Suno at all; it settled instead with rival Udio and remains in litigation with Suno in court. Apple Music, meanwhile, began rolling out a “Made With AI” label based on artist self-disclosure through its Transparency Tags framework. And on July 10, a coalition including the RIAA, IFPI, the Grammys, and SAG-AFTRA launched a two-tier labeling standard: a bold black “AI” block for a recording generated primarily by AI, a smaller white “ai” for one made substantially by humans where AI touched only some expressive elements. That’s what changes the conversation from “is this allowed” to “who gets paid, for what, and how do we even tell the difference.”

I keep coming back to one question underneath all of it: when a machine helps make the thing, whose hand is actually on the instrument?

The tool that talks back

A paintbrush doesn’t suggest a color. A guitar doesn’t propose a chord change. But the newest generation of creative AI does — it listens to what you’ve made so far and offers a direction, sometimes a good one. That’s a different relationship than “tool.” It’s closer to a collaborator who never gets tired, never has an off day, and has no stake in whether the work is honest.

Most musicians I’ve read on this don’t want AI to replace the work — they want it to clear the underbrush. One survey this year put overall AI adoption among musicians and producers at 87%, but the breakdown tells the real story: over half are pointing it at the parts of the job nobody romanticizes — cover art, bios, captions, the endless promotional grind. Not the melody. Not the lyric that took three years to get right. The chores around the art, not the art itself.

That’s a reasonable line to draw. Whether it holds is another question — the tools keep getting better at exactly the parts people swore they’d never hand over. And listeners are pushing back harder than the industry expected: a Luminate study tracking attitudes from May to November 2025 found overall interest in AI music sliding from -13% to -20%, and on Deezer, AI-made tracks now account for 44% of daily uploads but under 3% of actual streams — most of those plays later traced back to bots, not people. Apple’s own numbers echo it from the other side: Apple Music VP Oliver Schusser has said more than a third of monthly uploads are now fully AI-generated, yet they draw less than 0.5% of actual listening. The resistance is sharpest among Gen Z and Gen Alpha — which is awkward, since they’re exactly who the platforms are courting with AI remix features. Volume isn’t the same as demand.

Authorship was never as clean as we pretended

Every era of music has borrowed its tools from wherever it could — the synthesizer was once accused of not being a “real” instrument, sampling was once theft before it was genre-defining, and the producer’s chair has always blurred the line between performer and author. AI collaboration isn’t the first time we’ve had to ask who made this, really. It might just be the first time the honest answer is: several people, a machine trained on thousands of other people’s work, and a human who chose what to keep.

What feels new is the scale and the speed — and the fact that the training data itself is someone’s uncredited labor. A label-endorsed model trained on a licensed catalog, like the Suno–BMG deal, is at least honest about that transaction. An unauthorized one just launders it.

What I think this means for people who make things by hand

For a small creative studio like this one — music, art, and stories, made mostly alone, mostly slowly — the temptation isn’t to compete with what AI can generate at scale. It’s to notice what it still can’t fake: a body of work with a history behind it, choices that trace back to a specific life, a voice that took years, not seconds, to sound like itself.

I don’t think the answer is refusal. I don’t think it’s surrender either. It’s closer to what the musicians surveyed this year keep saying: use it like a tool, not a shortcut. Let it clear the underbrush. Keep your hand on the part that has to be yours.

The instrument changed. The question of whose hand is on it hasn’t.

If you make things by hand too — music, art, stories, or something in between — I’d like to know where you draw your own line. That’s what this corner of AbeArtsSweden is for.


← Back to Music