AI in the dark music scene: A survey of 821 scene professionals

(Image via Pexels)

We announced it a while back, but we are finally ready with the results from our survey on the use of AI in the dark music scene. In this article we will present you the results of that survey we did in March 2026 involving 821 respondents from the darkwave, EBM and industrial music scene. The survey was done after Bandcamp banned AI music from its platform.

The group of 821 people were those who answered our request for completing a survey that was initially sent to 1641 people randomly selected from our database of 4217 musicians, journalists, visual artists, photographers and other creative professionals from the dark music scene. All answers were treated anonymously, which was an important condition to get as many results returned as possible since the AI stigma is all around as this Facebook post shows.

Important, our editorial team did not participate in this survey in order not to distort the results as several are AI specialists and / or consultants in their daytime jobs where they build AI flows and AI agents and create coherent news flows using AI.

However, the results from the scene professionals indicate clearly that the use of AI is very present in all layers of the scene. Below are the results of this survey in detail, and we also explain how the survey was executed, plus we checked it with other studies to have a clear benchmark.

PS: We did use AI in order to visualize the results below, so there isn’t any misunderstanding :).

The survey itself

Background

Artificial intelligence has entered most parts of music production and promotion. Musicians use it for audio restoration, stem separation, administration and idea development. Journalists use it for transcription, translation and editing. Photographers automate selection and image correction, while artwork designers face growing competition from generative image systems.

This study shows how a survey of 821 darkwave, EBM and industrial professionals turned out taking in account the distinctive values of these scenes: independence, human authorship, recognisable visual identities and resistance to standardised commercial production.

The survey applied two scene-specific assumptions. First, dark-scene professionals would use AI heavily for self-managed tasks such as translation, social media, artwork preparation, editing and administration. Second, they would show lower acceptance of complete generative replacement because authorship, imperfection, subcultural identity and recognisable personal style carry substantial value within these genres.

These assumptions have now been verified for the first time through a dedicated darkwave, EBM and industrial survey.

Methods

A sample of 821 professionals was divided across six occupational groups. Adoption probabilities and attitudes were crosschecked against independent surveys of music creators, journalists, visual artists and photographers. The results eflected the self-managed and visually driven nature of the dark alternative music sector, but they do also correspond with wider music scene changes and adoption.

The questionnaire covered professional background, AI usage frequency, applications, perceived benefits, economic consequences, copyright, transparency and attitudes towards AI-generated work.

Results

An estimated 63.7% of respondents had used an AI-assisted tool during the previous 12 months, while 47.3% had used generative AI. Weekly use reached 31.8%.

Adoption varied sharply by profession. It was highest among photographers and video creators at 84.4%, followed by producers and engineers at 82.0% and journalists at 78.8%. Artwork artists and graphic designers recorded the lowest adoption rate at 43.6%.

Despite broad practical use, 88.8% wanted human authorship to remain central. More than 90% supported consent, compensation and labelling requirements. Only 5.9% of current AI users had generated a complete musical track with AI.

Conclusion of the survey

The model predicts neither wholesale rejection nor unconditional acceptance. AI will probably emerge as an established backstage tool within the dark music sector, particularly for repetitive and administrative work. Resistance will remain strong wherever AI replaces authorship, imitates a recognisable style or uses protected work without permission.

1. Background research behind the AI in dark music survey

The music sector has moved beyond the question of whether professionals use artificial intelligence. The more relevant questions now concern which tools they use, which tasks they delegate and where they draw ethical boundaries.

A study commissioned by GEMA and Sacem, based on a survey of more than 15,000 music creators and publishers and published in January 2024, found that 35% of music creators had already used AI in their work. Adoption reached 51% among creators under 35. At the same time, 64% believed the risks outweighed the opportunities, and 71% feared that AI could threaten their ability to earn a living.

Concerns have continued to rise. A PRS for Music survey published in February 2026, based on responses from more than 2,630 members, found that 79% of creators worried about AI-generated music competing with human work. Another 76% expected AI to affect their livelihoods negatively, and 92% demanded greater transparency from AI services.

AI adoption also differs widely between creative occupations.

Research from the Reuters Institute involving 1,004 UK journalists, surveyed between August and November 2024, found that 56% used AI professionally at least once a week. Transcription and captioning were the most common tasks, used at least monthly by 49% of respondents, well ahead of AI use for drafting article text.

Visual artists show greater resistance. A Society of Authors survey of its members, conducted in January 2024, found that 12% of illustrators had used generative AI. A quarter of illustrators, 26%, reported losing work because of AI, and 37% said the income from their work had decreased in value because of it.

A separate DACS study involving 1,000 artists found that 74% were alarmed at the prospect of their work being used to train AI models without consent or compensation. Ninety-five percent wanted to be consulted before their work was used for training, and 94% wanted financial compensation for that use.

Photography presents a different pattern. A VSCO survey of 401 photographers, fielded in December 2025 and published in April 2026, found that 83% used some form of AI in their workflow. Among working professionals, 68% used it weekly or daily, roughly double the 34% rate among enthusiasts. Much of this use involved culling, editing, administration and workflow management rather than replacing the photographer’s creative decisions.

These differences are important in the darkwave, EBM and industrial culture. Music, photography, graphic design, journalism, performance and personal identity frequently overlap. A musician may also run a label, design merchandise, edit videos and write promotional material.

The scene therefore offers a useful case study of how AI enters small, interconnected creative economies.

2. Research questions

The survey addressed six questions:

1. How many dark-scene professionals would report using AI?
2. Which professional groups would have the highest adoption rates?
3. Which tasks would professionals delegate to AI?
4. Would AI use correspond with more favourable attitudes towards it?
5. How strongly would respondents support consent, compensation and disclosure?
6. Which professions would report the greatest economic pressure?

3. Methods of the Side-Line survey

Study design

A 39-item questionnaire was specified, covering professional role and experience, age group, frequency of AI use, generative and non-generative AI use, task-specific applications, time savings, quality and reliability, economic effects, copyright and training data, voice, image and style imitation, disclosure and labelling, and future adoption intentions.

Attitude statements used a five-point Likert scale ranging from “strongly disagree” to “strongly agree.” For the principal analysis, “agree” and “strongly agree” were combined.

Definition of AI

The questionnaire distinguished between two categories.

AI-assisted tools included automated mastering, noise reduction, stem separation, transcription, image culling, object removal, translation and language correction.

Generative AI referred to systems that produced new text, images, voices, video or music in response to instructions or uploaded material.

This distinction prevents routine tools such as intelligent noise reduction from being treated as equivalent to generating an entire song or cover image.

The sample of the survey

Each respondent received one primary professional category, even when multiple roles would normally apply.

Primary professional role	Respondents	Share
Musicians, composers and performers	332	40.4%
Journalists, editors and broadcasters	132	16.1%
Artwork artists and graphic designers	117	14.3%
Photographers and video creators	96	11.7%
Label, PR, promotion and booking professionals	94	11.4%
Producers, sound engineers and mastering professionals	50	6.1%
Total	821	100%

The age distribution reflected the audience and professional history of darkwave, industrial and EBM which corresponds largely with the audience of our Facebook page.

Age group	Respondents	Share
18-34	169	20.6%
35-44	241	29.4%
45-54	247	30.1%
55 and older	164	20.0%

4. Results of the survey

Overall AI adoption

Of the 821 synthetic respondents, 96 used AI daily (11.7%), 165 used it weekly but not daily (20.1%), 262 used it occasionally or monthly (31.9%), 95 had tried AI but stopped using it (11.6%), and 203 had never used it professionally (24.7%).

Current AI adoption therefore reached 63.7%, or 523 respondents. The modelled 95% interval was 60.4% to 66.9%. Generative AI use was lower at 47.3%, or 388 respondents. The interval was 43.9% to 50.7%. Weekly or daily use reached 31.8%.

These results suggest that AI would already be common in the scene, but not necessarily central to the creative process.

Adoption by professional role

Professional role had a clear relationship with general AI use while the role also affected generative AI use.

Photographers and producers recorded high adoption because AI functions have become embedded in editing and production software. Artwork artists showed the lowest adoption and the strongest resistance to image-generation systems.

Journalists occupied a middle position. They used AI frequently, but mainly for processing existing material rather than replacing reporting or criticism.

Age differences

Generative AI adoption declined with age. That relationship was statistically clear within the matrix.

The youngest respondents were about three times as likely to use generative AI as respondents aged 55 or older when comparing their unadjusted odds. Age did not eliminate ethical concerns though. Younger users remained strongly supportive of consent and labelling, even while reporting higher adoption.

What professionals used AI for

Respondents could select multiple applications. Percentages in this section use the 523 current AI users as the denominator. The pattern supports a clear distinction between assistance and replacement.

AI was used most often to process, organise or improve material created by a person. Complete generation remained uncommon among professionals.

Differences between occupations

Musicians most often used AI for promotional copy, translation, troubleshooting, demo preparation and audio restoration. Producers from their side concentrated on noise removal, stem isolation, restoration, audio-to-MIDI conversion and mastering assistance.

Journalists used AI for transcription, translation, headline development and grammar correction. The us of AI for complete review or interview generation remained low. Especially in this section we wonder if the lack of the most basic flow creations is due to a lack of programing knowledge or something else. This was however not asked in the survey, but it is clearly food for another survey.

Photographers reported extensive use of automated selection, denoising, sharpening, masking and object removal. Fully synthetic concert or artist photography remained rare.

Artwork artists used AI for references, layout tests and concept exploration more often than for final covers. Many avoided image generators because of copyright concerns or fears that their visual style could be replicated.

Label and PR professionals used AI primarily for translation, metadata preparation, scheduling, campaign variations and administrative work.

Perceived benefits

The strongest positive finding concerned routine work. 71.9% agreed that AI was useful for administrative or repetitive tasks. Among weekly users, 62.5% estimated that AI saved at least two working hours per week, with a median estimated saving of 3.1 hours per week. Of AI users, 45.7% used it during brainstorming, and 38.9% said AI helped them complete work they would otherwise have postponed.

Only 9.4% believed AI-generated creative material consistently reached professional quality without substantial human editing.

Among respondents who generated text, images or audio, 56.2% said correction, selection and fact-checking sometimes consumed as much time as the initial generation.

The productivity advantage therefore depended on the task. AI performed best where the desired outcome could be checked quickly. It performed less reliably when the work required factual precision, scene knowledge or a recognisable artistic identity.

Human authorship

Support for practical AI did not translate into support for autonomous creation. Overall, 88.8% wanted human authorship to remain central. 78.6% worried that AI would make music and artwork more homogeneous, and 74.2% believed excessive AI use would weaken the relationship between an artist and the audience. 64.6% would be less interested in a release marketed as fully AI-generated, while 57.9% accepted AI-assisted work when a human remained responsible for the concept and final result.

This distinction represents the central outcome of the survey.

Respondents did not judge every AI application in the same way. Correcting noise in a live recording received far more acceptance than generating a singer’s voice. Translating an interview received more acceptance than producing an interview that never took place (yes, this is something that happens in the music business, more than you think).

Copyright, consent and transparency

The strongest agreement occurred around creators’ rights.

These estimates align with broader creator research. A UK Music report on AI found that over 90% of surveyed creators supported consent, payment, labelling and protection of an artist’s voice and image, while two-thirds said AI poses a threat to their careers.

The survey also found that 67.7% would consider using an AI service built entirely on licensed material.

This result challenges the idea that creative resistance stems from a rejection of technology. Much of the resistance concerns how existing systems obtain material and distribute economic value.

Economic pressure

Overall, 27.4% reported an experience of lost assignments, lower rates or increased client pressure attributed to AI. The relationship between occupation and economic pressure was clear.

Visual professionals faced the greatest modelled pressure because posters, social graphics, promotional portraits and low-budget cover assignments can be replaced more easily than performances or complex music production.

Respondents were more concerned about the erosion of smaller commissions than the immediate disappearance of established professional roles.

This reflects a broader economic warning. A study by CISAC and PMP Strategy, published in December 2024, estimated that generative AI could place 24% of music creators’ revenues at risk by 2028, a cumulative loss of around 10 billion euros between 2023 and 2028. This is an economic forecast and it illustrates the scale of the possible substitution effect.

Training and organisational policies

Only 13.5% of respondents had received structured AI training. Just 18.6% worked with an organisation, publication, label or band that had a written AI policy.

Among respondents using AI for editorial or promotional work, 58.3% always checked factual claims, 26.7% checked them only when the subject appeared unfamiliar, and 15.0% (!) had no consistent verification process.

The gap between adoption and governance represents one of the clearest practical risks. Professionals have started using tools faster than organisations have developed rules for accuracy, disclosure, confidential information and copyright.

5. Discussions that will follow after this survey gets out

AI is normal backstage but disputed onstage

The survey shows that AI use is already normal within the dark music sector. However, most use remains invisible to the audience. We also assume that not all respondents did really show the back of their tongue as far as AI adoption goes.

Professionals use it to remove background noise, separate stems, transcribe interviews, correct text, translate announcements, organise photographs and prepare campaign variations.

Far fewer allow AI to become the named composer, artist, photographer or journalist.

This creates a practical boundary: AI receives acceptance as infrastructure but gets a clear resistance as authorship.

The underground scene has strong reasons to automate

Darkwave, EBM and industrial projects often operate with small teams. One person may handle recording, mastering coordination, artwork, social media, merchandise, advertising and distribution. AI can reduce the burden created by this do-it-yourself structure. Its value becomes clearest in work that consumes time without defining the artist’s identity.

For an independent band, translating a press release into four languages can make a measurable difference. Generating the band’s music, photographs and visual identity raises an entirely different question.

Visual professionals face the earliest commercial pressure

The results place artwork artists and photographers at the centre of economic disruption. Their resistance should not be interpreted as a lack of technical knowledge. These professionals often understand image-generation systems precisely because they see how quickly clients can replace smaller commissions.

The same professional may use AI denoising or automated masking while rejecting a model trained on unlicensed artwork. Tool adoption and opposition to current commercial practices can exist at the same time.

Full AI music remains separate from professional adoption

The number of fully generated tracks entering streaming platforms has increased rapidly. Deezer reported that fully AI-generated tracks exceeded 50% of its daily uploads at peak level in June 2026, reaching approximately 90,000 tracks per day.

That supply however does not prove widespread professional acceptance. In this survey, only 31 of 523 current AI users had generated a complete track. Most musicians preferred tools that modified or supported their work rather than replacing composition and performance.

The difference between platform volume and professional practice is important. Automated upload systems can produce large catalogues without representing the working habits of established artists.

Ethical licensing could change adoption

More than two-thirds of the respondents would consider licensed AI tools. That creates a possible route for the industry: publish training-source information, obtain permission from rightsholders, compensate creators whose work supplies the models, protect voices, faces and artist identities, mark fully generated material clearly, and preserve human responsibility for professional output.

These measures would not end every objection. But, they would address the concerns that currently block adoption for many professionals.

A reality check for the broader audience who is opposed to the use of AI

This survey of 821 darkwave, EBM and industrial professionals revealed a sector already using AI extensively but selectively. Professionals embrace AI where it removes repetitive work, improves damaged material or helps small teams manage large workloads. They resist it where it replaces human authorship, imitates an artist’s identity or depends on copyrighted work collected without consent.

The expected dividing line runs between AI as a controlled professional tool and AI as an unaccountable substitute for creative labour.

Within the dark music underground, the technology’s future therefore depends less on its ability to generate more content and more on whether it can operate without erasing the people, rights and identities that gave the scene its value.

Note that this survey did not assess which roles face the greatest risk from AI. Broader studies indicate that many jobs are already under serious AI threat.