Gisteren (2 juni 2026) was ik in Brussel, en sprak ik een delegatie Europarlementariërs en adviseurs van de Europese Commissie toe over de gevolgen van generative AI voor muzikaal auteursrecht. Hieronder is mijn speech te lezen. De conferentie die ik mocht afsluiten, werd georganiseerd door ECSA, de Europese koepelorganisatie voor componisten en songwriters.
Good afternoon, thank you so much for sticking around, all of you here in the room who are longing for a drink as much as I do, but of course also to the people joining us online. Let me start off with thanking you again Mr. Voss and Mr. Sousa Silva for your inspiring speeches.
My name is Aafke Romeijn, I am a Dutch songwriter and musician and a member of the boards of both ECSA and BAM! Popauteurs, the largest advocacy organisation in the Netherlands for creators of pop music, hip-hop and dance music.
About a year ago, I woke up in the morning and I saw I got a few very strange messages from friends and fans. They all read roughly the same:
“Hey Aafke, congrats on your new single you released today, we hope it’ll get a lot of streams, but we’re also a bit concerned, because it doesn’t really sound like you. It sounds a bit weird. Are you ok?”
I was very surprised. Not because people thought my music sounded weird, that happens more often than not, but because I hadn’t released a single that day. At least: not that I knew of. So I went to my Spotify Artist profile, and lo and behold, there it was: a brand new single, released by Aafke Romeijn. I wondered if I somehow had recorded and released a song while I was asleep.
But then I started listening. Nope, that definitely wasn’t me, not even a sleepwalking version of me. It sounded like a crazy uptempo polka song with the weirdest dutch lyrics ever, something about me buying a horse at a flea market. Now I am many things, but certainly not a horse girl.
I had no idea what had happened. As far as I knew, the only people that could release music on my Spotify Artists profile were me and my record label. So I called them frantically and asked them: What on earth have you put on my page? To my surprise, they had no idea either, they listened to the doomscroll-polka-brainrot and said: “Ah, this must be AI.”
Of course it was AI. What else sounds so crappy?
What scared me was not just the low quality of the AI song, but maybe even more that someone could just impersonate me and release music on my own profile. The only way to solve this was to file a complaint with Spotify and hope they delete the song quickly enough. In the meantime, thousands of my listeners have heard “me” sing about buying a horse at a flea market, which, frankly, is terrifying.
And that is what we have addressed today as well: AI is everywhere, in every crevice of our artistic identity, it threatens not only our income or our status as artists, but also our sense of self and identity.
Of course, AI’s impact on authors’ rights remains the most urgent issue to address. As said by Helienne earlier, the damage has already been done: unlicensed AI services have scraped and ingested all of our songs and creative input before we even had the time to pronounce the words “opt out”.
And as stated in the report of Mr. Voss, we must find a solution to ensure that generative AI respects the fundamental principles of consent, transparency and fair remuneration. Transparency is an essential pillar for both consent and remuneration, as well as the basis on which we can build a functioning licensing market. We want to know what we’re listening to, and we want artists to get a fair share.
Today, we as music creators find ourselves cornered. Billion-dollar AI companies like Suno exploit our work without our authorisation, without transparency or remuneration. You may think that the recent licensing deals between the major labels and some of these services are a positive step. Probably, yes, but please notice this: they are only signing deals with those who have the power to sue them, leaving individual music authors in the dark about the details of these contracts. In this context, it is more important than ever that we ensure that licensing can benefit all of us music creators, and not just the largest players.
But, as we have heard today, and as I’ve experienced firsthand: the effects of AI go well beyond copyright debates, they’re touching every aspect of our profession.
The massive proliferation of AI-generated music on music streaming platforms is already reducing royalty payments to real creators by diluting the finite revenue pool generated by streaming. Each cent that goes to an AI-generated song, even if it’s a better song than the polka rubbish on my artist page, is a cent that is being taken away from real songwriters, performers and other rightsholders.
It is also clear that most AI-generated music is made for the sole purpose of committing streaming fraud, such as what happened on my own streaming profiles. The mechanism is simple: fraudsters use free and unlicensed services to generate thousands of fake songs. They upload to streaming platforms, then use bots to stream them and take royalties away from real creators and rightsholders. The numbers speak clearly: in April, Deezer reported that 85% of streams generated by fully AI-generated music tracks are fraudulent.
When AI-music was released via my artist page, I quickly discovered that loads of other dutch artists had had the same issue – it was a violation of my personality and authors’ rights and outright fraud. Unfortunately, by the time you report this to the streaming platform and the AI content gets removed, it has stayed on the profile for hours or days, collecting streams and deceiving listeners. And it is not just happening in pop music alone: it’s a phenomenon that has been targeting a lot of successful artists across genres, including very famous artists in the jazz scene.
As stated in the Voss report and in the 2024 report by the European Parliament, we need to take bold actions against deepfakes and impersonation – this is not just an issue for creators, but for each one of us who has an online presence. And every actor in the value chain has a role to play to end streaming fraud, in particular distributors and streaming platforms, who should verify the identities of both content uploaders and users to ensure they are real people, not just bots generating artificial streams. And fight against suspicious accounts and block fraudulent streaming activity.
Alongside tackling royalty dilution, streaming fraud and AI impersonation, ensuring transparency across the value chain is also needed. Music industry stakeholders must work together on finding effective remedies to the challenges posed by the proliferation of fully AI-generated music. Our second panel discussed possible solutions to this, from the use of AI detection tools and the improvement of music metadata, to obligations for digital streaming platforms to label AI-generated music.
Lastly, as we have heard in the presentation of Mr. Clarke, digital streaming platforms being flooded with synthetic content risks making it harder for authentic, human-created music to be discovered. The findings of the European Commission study on discoverability are powerful and point to the need to take action – from promoting transparency of recommendation systems, to ensuring the prominence of European works and the authenticity of content in the era of generative AI. Mr. Sousa Silva’s upcoming report on AI in cultural and creative sectors can play an important role in calling on the European Commission to put forward bold measures on these fronts.
Thank you for your attention. I would like to thank very much again, on behalf of ECSA, all panellists and speakers for their insightful contributions to today’s discussions.
Our sincere thank you in particular to Executive Vice-President Henna Virkkunen for her opening keynote and to Members of the European Parliament Emma Rafowicz, Axel Voss and Hélder Sousa Silva for being here, but also for all their work and unwavering support of European music creators.
Thank you all very much for joining us here in person and online. And one last piece of advice that I want to send you home with: if one day you hear an artist you like release a new single that sounds a bit off, please send them a message. Cause you never know: it might not be them at all.
