AI Ate The Dubplate? Why Jungle & DNB Artists Are Worried About Music Being Used To Train Machines

AI Ate The Dubplate? Why Jungle & DNB Artists Are Worried About Music Being Used To Train Machines. As AI companies face lawsuits over music training data, JDNB looks at what this means for jungle, drum and bass, dubplates, underground artists and human-led music culture. The underground music world is watching the AI industry very closely right now and not in a calm way. Artists all over my feeds are increasingly concerned about AI being trained on their music without their knowledge, consent, credit, or compensation. For many musicians, a track is not just data; it represents years of practice, creativity, personal experiences, financial investment and artistic identity. When AI systems analyse millions of songs to learn patterns, styles, melodies, arrangements, and production techniques, some artists feel their work is being used to help create competing music products without permission. This can leave creators feeling powerless and worried about the future value of their craft.

While many artists accept that AI is becoming part of the industry, they want transparency, licensing, and fair payment. Musicians can take practical steps by reviewing distribution and licensing agreements, including clauses that prohibit AI training without written consent, registering their copyrights, joining industry organisations that advocate for artists’ rights, and supporting campaigns for stronger regulation. They can also continue to build their unique artistic voice, as originality, human experience, live performance, and genuine connection with audiences remain qualities that AI cannot fully replicate.

I'm going to take a look at what's happening, what can and can't be done.

For years, jungle and drum & bass has lived through sampling, remixing, dubplates, bootlegs, pirate radio culture, SoundCloud uploads, white labels, rare vocal cuts and scene-led creativity. But the new AI music debate feels different. This is not just about inspiration. This is about whether entire catalogues, independent tracks, metadata, moods, genres, tags, vocals, basslines and production fingerprints are being absorbed into machine-learning systems without clear consent, credit or payment. Can we also be hypocrites while most of the back cat of jungle and dnb has been pillaged?

The Amen break came from The Winstons’ 1969 track Amen, Brother... a seven-second drum break played by Gregory Coleman that helped shape thousands of jungle and drum & bass tracks, while the original musicians received no royalties from its widespread use. A crowd fund was done recently....

The panic around AI using music to train machines is real, and for underground scenes like jungle and drum and bass, it hits a nerve. Artists are asking fair questions: who gave permission, who gets paid, and what happens when years of sound design, culture, vocals, breaks and basslines are fed into a system that can then imitate the surface of the music without understanding the life behind it?

But there is an uncomfortable question too: can we also be hypocrites while most of the back cat of jungle and DNB has been pillaged? Jungle itself was built from sampling, chopping, reworking, borrowing and transforming sound. Amen breaks, reggae vocals, soul snippets, hip hop drums, dub sirens and rare groove fragments all helped shape the culture. The difference, though, is context. Sampling in jungle was often creative, risky, human and transformative. It came from DJs, producers and sound system culture pushing boundaries with emotion and intent.

AI training feels different because it can happen invisibly, at scale, and without relationship to the scene. So maybe the answer is not to deny our history, but to be honest about it: jungle has always recycled sound, but it has also given it soul. The machines still owe the culture respect.

The latest flashpoint is a lawsuit from Jamendo against Nvidia, where the independent music platform has accused the tech giant of using hundreds of thousands of audio files and related metadata to train AI audio systems. Whether the courts agree or not, the story has landed hard because it touches a raw nerve for independent artists: if underground music is being used to build commercial AI tools, who gets paid, who gets credited, and who even knows it happened? You can read the Jamendo story here.

This matters to jungle and drum & bass because the scene has always been built by people uploading, sharing, testing, remixing and pushing sound forward. A lot of underground artists do not have major-label legal teams. They may have music on YouTube, Spotify, SoundCloud, Bandcamp, Free Music Archive-style platforms, radio archives, old mixes, label channels and promo pools. That visibility is meant to help artists reach people. But in the AI age, some musicians now worry that being discoverable also means being scrapeable. So do artists lose rights once your music goes on a public platform? Having music on YouTube, Spotify, SoundCloud, Bandcamp or Free Music Archive-style platforms does not automatically mean an artist loses their rights. Copyright usually stays with the creator unless they sign it away. But the risk is more subtle: artists often grant broad platform licences so their music can be hosted, streamed, shared, recommended, embedded or promoted. That does not always mean theft, but it does mean control can become blurred.

So the question is: if underground artists upload music to free or public platforms to be heard, are they also making themselves easier to scrape, copy, analyse or feed into AI systems without fully understanding what happens next?

And can we also be hypocrites while most of the back cat of jungle and DNB has been pillaged? Jungle was built on sampling, chopping and reworking sound, but usually by humans transforming records into something raw, emotional and culturally alive. AI training feels different because it can happen invisibly, at scale, and without a relationship to the scene.

Artists are not always losing ownership overnight. They are losing certainty, leverage and visibility over how their work is being used. That is why consent, credit and payment matter now.

The fear is not simply “AI exists.” Many producers already use technology, plugins, stem tools, mastering assistants and creative software. Jungle and DNB have never been afraid of machines. The culture itself was built on samplers, sequencers, chopped breaks and studio experimentation. The real issue is consent. There is a big difference between using tools to make music and using someone else’s life’s work to train a system that can then compete with them.

For underground artists, the anxiety is even sharper. If an AI model learns from major artists, at least those artists may have lawyers, publishers and labels watching. But what about the independent DNB producer with 700 monthly listeners? What about the jungle vocalist who uploaded a hook ten years ago? What about the small label that spent years building a sound? What about the producer whose drums, bass design, ragga vocal style or atmospheric pads become part of a model’s “understanding” of jungle music?

This is why the conversation is moving from “AI is cool” to “who trained it, what did they use, and who gave permission?”

The Atlantic’s recent AI music reporting has intensified the debate, with artists checking whether their songs appear in datasets linked to AI music generation. The Atlantic “The Millions of Songs Mashed Into AI-Generated Music” What was this? It’s an investigation by Alex Reisner into four huge music datasets being shared around AI research/development circles. The Atlantic says these datasets collectively include more than 21 million tracks, with music linked from places like YouTube, Spotify and Free Music Archive-style sources. It also launched an AI Watchdog search tool where artists can look up whether their tracks appear in those datasets. AI Watchdog investigation has added fuel to the fire by making searchable several huge music datasets circulating around AI development. According to The Atlantic, these datasets contain more than 21 million tracks, pulling from platforms and archives that include YouTube, Spotify and Free Music Archive-style sources. The issue is not just that music exists online; it is that artists can now search and discover their work may have been swept into training-data pipelines without clear consent, credit or payment. For underground jungle and D&B artists, that raises a serious question: if being discoverable online also makes your catalogue easier to scrape, who is actually protecting the smaller creators? Read more here. Some musicians have reacted with anger, especially where they feel their work has been taken into systems that may eventually generate music in similar emotional, sonic or cultural lanes. For dance music, this raises a huge question: can a machine replicate the sound without understanding the scene? I don't think so at all. I think labels can do more. We can do more. What can labels actually do? Underground labels may need to start treating AI checks like sample clearance, not paranoia. If a track is submitted, labels can ask for stems, project files, rough exports, session screenshots, sample sources and a simple written declaration of whether AI was used. Not to shame artists, but to protect everyone involved.

There is also new test that I do: put suspicious music into Suno-style tools and see whether the system accepts it, extends it, eats it up and regurgitates something that feels too close. That might become a useful red flag, especially if a track behaves like it was born inside the same machine. But it should not be treated as final proof. AI detection is messy, platforms change, and false positives could hurt real artists. I tested this this morning. I uploaded a free track "Listen" into Suno which was made completely from scratch as a remix and is up for a free download. It "detected some interesting sounds" which I thought was funny, it's probably now going to use these interesting sounds it's never come across. While it's a free download, never gone through Beatport, Juno or Apple it still refused to use it. Did it analysise my "interesting sounds?" Who knows.

Interestingly I then put my new release in "On The Bus" and it also "detected some interesting sounds" and then I had the same error.

I have interestingly put in some other tracks from other producers and it has accepted them which makes me wonder, was that made using AI?

The bigger point to me is authenticity and disclosure. Was the tune built by a producer, with human choices, arrangement, mix decisions, sound design and intention? Or was it generated, dressed up and passed off as fully human-made? Personally if a piece of music is good and it's been disclosed as AI, great. It's the falsehood that bothers me.

Labels can protect the culture by asking better questions before release. Not “are you using tools?” but “how was this made, who owns every part of it, and can we prove the creative chain?” A simple 2 minute video of the artist infront of a DAW. I was accused of having my music ghost produced, kinda a compliment in a way so I made my entire last album live on facebook live videos and instagram live videos from scratch. Maybe we need more of this? Maybe fans need to get a bit more engrossed in "how did you make this?" rather than "yeh sick track" over the 50 uploads a month of "new" music which let's face it, is physically nearly impossible. It begs the question around throwaway music too?

Jungle and drum & bass to me is not just tempo and drums. It has a sound. You can hear the sound from established veterans vs popular commercial artists who have jumped ship temporarily to make a few quid. The are culture, history, Black music influence, sound system energy, pirate radio, MCs, dubplate pressure, rave sweat, label families, regional scenes, rewinds and lived experience is all part of what we breath. AI may be able to imitate surface-level ingredients, I still don't think it really can imitate a proper crusty hand made amen break. AI reese bass, can be identified by the wav form in processing. AI chopped vocals, sound very "clean" and ragga phrases, easy to spot as there are no credits to the original vocalist. So should labels start insisting on this? It's certainly a thought right? Should people be putting out music with ragga AI vocals and not reaching out to our amazing talented vocalists and MCs if they really care about our underground community? Liquid pads, neuro growls, the lists go on... but the underground is not only a collection of sounds. It is a network of people. Labels don’t need to become anti-technology they just need to become pro-provenance imo. And if that is a bit too big wordy for you, record labels don’t need to be against AI or new technology. They just need to care more about proving where music/content comes from.

That is why JDNB believes this conversation needs to stay human-led. The answer is not blind panic, but it is not passive acceptance either. Artists need transparency. Labels need clarity. Platforms need to explain what is being used. AI companies need proper licensing models. And underground musicians need a seat at the table before their work becomes fuel for tools they never agreed to build. Labels need to do some homework and artists, if they really care need to be releasing on labels that also care.

There is also a danger that the lowest-effort AI music floods streaming platforms and weakens discovery for real producers. If thousands of AI-generated “jungle” or “drum and bass” tracks can be made instantly, then human artists are not only competing with each other, they are competing with machines trained on the culture itself. That could make it harder for listeners to find real underground music, and harder for small artists to build sustainable careers.

There is still power in our scene. Jungle and DNB have survived industry neglect, radio bans, (the scene was heavily carried by pirate radio stations like Kool FM, Rinse, Rude, Flex FM and others, because mainstream radio often ignored, misunderstood, or under-supported it). Commercial waves, underground splits, streaming chaos and every trend that claimed it would replace real culture. The strength of this music has always been its people. DJs, producers, MCs, promoters, ravers, writers, radio hosts and labels are the reason the sound keeps mutating and surviving along with the community.

So maybe the question is not “Can AI make jungle?” Maybe the real question is: can AI ever belong to jungle without respecting the people who built it?

For now, the message from our underground heads should be clear: innovation is welcome, theft is not. (Does beg the question about bootlegs..... is that hypocrisy? Maybe a discussion for another day!) Tools are welcome, exploitation is not. Technology can support creativity, but it should not erase the creators.

JDNB will keep watching this closely because if the machines are learning from the underground, the underground deserves to know. Written by Missrepresent 24.06.26