Towards Music Industry 5.

0: Perspectives on Artificial
Intelligence
Alexander Williams, Mathieu Barthet

To cite this version:

Alexander Williams, Mathieu Barthet. Towards Music Industry 5.0: Perspectives on Artificial Intel-
ligence. Workshop on AI for Music, Association for the Advancement of Artificial Intelligence, Mar
2025, Philadelphia (USA), United States. �hal-04943901�

HAL Id: hal-04943901

https://hal.science/hal-04943901v1
Submitted on 12 Feb 2025

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Copyright
 Towards Music Industry 5.0: Perspectives on Artificial Intelligence

Alexander Williams1 , Mathieu Barthet1,2

1
Centre for Digital Music, Queen Mary University of London
2
Aix-Marseille Univ CNRS PRISM
[email protected], [email protected]

Abstract The music tech sector typically seeks to develop disruptive

or value-adding technologies that challenge the industry sta-
Artificial Intelligence (AI) is a disruptive technology that is
transforming many industries including the music industry. tus quo in areas such as music production, publishing, con-
Recently, the concept of Industry 5.0. has been proposed em- sumption, or distribution, while music rights-holders seek to
phasising principles of sustainability, resilience, and human- protect the value of their intellectual property (IP) - primar-
centricity to address current shortcomings in Industry 4.0. and ily music recordings and artist brands. In many cases, the
its associated technologies, including AI. In line with these success and survival of an emerging music tech company de-
principles, this paper puts forward a position for ethical AI pends upon securing licencing arrangements for copyrighted
practices in the music industry. We outline the current state material from asset holders or obtaining support through
of AI in the music industry and its wider ethical and legal partnerships. This power dynamic means many emerging
issues through an analysis and discussion of contemporary music tech companies will be resisted if they are seen not
case studies. We list current commercial applications of AI
to align with the corporate strategic goals of the incumbent
in music, collect a range of perspectives on AI in the indus-
try from diverse stakeholders, and comment on existing and oligopoly (Watson and Leyshon 2022).
forthcoming regulatory frameworks and industry initiatives. Current music tech innovation is rooted in the maturity of
Resultingly, we provide several timely research directions, Industry 4.0 technologies such as AI, big data, cloud com-
practical recommendations, and commercial opportunities to puting, virtual / augmented reality (VR / AR), the Internet
aid the transition to a human-centric, resilient, and sustainable of Things, and blockchain. Such technologies are (socially)
music industry 5.0. This work particularly focuses on west- disruptive (Hopster 2021) and are currently transforming the
ern music industry case studies in the European Union (EU), music industry by introducing new business models, rev-
United States of America (US), and United Kingdom (UK), enue streams, and methods of music distribution and en-
but many of the issues raised are universal. While this work gagement (Jahromi and Ghazinoory 2023; Mbamba 2024;
is not exhaustive, we nevertheless hope it guides researchers,
businesses, and policy makers to develop responsible frame-
Clancy 2021). In 2021, the European Commission proposed
works for deploying and regulating AI in the music industry. the concept of Industry 5.0. Where Industry 4.0 focused on
increasing production efficiency, flexibility and worker up-
skilling through technical innovation, Industry 5.0 focuses
Introduction on using the encompassed technologies to achieve societal
Cultural and creative industries produce and disseminate goals beyond jobs and growth, such as social fairness, sus-
cultural products influenced by people’s lifestyles, beliefs, tainability, and worker wellbeing (Xu et al. 2021).
attitudes, and insights (Jahromi and Ghazinoory 2023). They Recent reports suggest that, without intervention, music
provide economic value and contribute to innovation, em- sector workers stand to lose nearly a quarter of their income
ployment, and national competitiveness (Mbamba 2024). to AI in the next four years (Cisac and PMP Strategy 2024)
Music is one such industry, and generally refers to individu- and up to 30% of (UK) jobs are automatable with AI with
als and organisations that earn money by creating and selling “crafts, creative arts and design” roles amongst those cur-
various forms of music, performing, and organisations and rently most at risk (Department for Education 2023). Given
professionals that aid, train, assist, represent and supply mu- the fundamental human connection in music (Malloch and
sic creators (UK Music 2024). Trevarthen 2018), it is essential that the music industry tran-
The makeup of the music industry is not static and has de- sitions to an Industry 5.0. model and adopts its three core
veloped closely in relation to technological innovation and principles of human-centricity, sustainability, and resilience
the creation of new music technology (music tech) (Lerch to ameliorate the existential challenges posed by AI.
2018). Some may consider music tech companies a segment
of the overall music industry, but tensions have always ex- Related Work
isted in their complex relationship with the wider industry. In recent years, there has been an increasing number of
Copyright © 2025, Association for the Advancement of Artificial works discussing ethical issues of AI in music applications
Intelligence (www.aaai.org). All rights reserved. generally and in specific commercial case studies, partic-
ularly generative music. (Mbamba 2024) reviews the im- cations of AI in the music industry will emerge with maturity
pact of Industry 4.0 technologies on global creative indus- in their technology readiness level (TRL).
tries, particularly focusing on AI for music creation and rec-
ommendation, VR / AR, and blockchain for digital rights Public and Professional Attitudes to AI in the
management. (Oğul 2024) contrasted ethical guidelines pub- Music Industry
lished by various AI researchers, music industry organi-
sations, and campaign groups. (Barnett 2023) conducted a AI has been described as a floating signifier in that it can
systematic literature review on ethical implications of gen- mean different things to different people (Suchman 2023).
erative audio models while (Jabour 2024) focused on the It has also been suggested that those with familiarity and
perceptions, ethical concerns, and business opportunities of expertise with AI are more likely to support its general ap-
AI-generated vocals. (Holzapfel, Sturm, and Coeckelbergh plication (Horowitz et al. 2024) and that there are numer-
2018; Huang et al. 2023) present ethical considerations relat- ous factors that can lead to incorrectly assessing AI’s capa-
ing to music information retrieval technology (MIR), while bilities (Crompton 2021). Therefore, any general opinions
(Peeters 2021) looks more closely at the impact of AI on on AI should be taken cautiously, particularly as it has cap-
MIR in general. (All-Party Parliamentary Group on Music tured mainstream public attention in recent years. Neverthe-
and UK Music 2024) presents a report on AI in relation to less, while opinions on AI in the music industry vary even
the UK music industry. (Huang, Sturm, and Holzapfel 2021) amongst similar stakeholders, common themes emerge.
presents an East-Asian ethical perspective on applying AI to
music applications. (Ma et al. 2024) present a survey of AI Artists and Music Industry Professionals
models for music applications with discussion of ethical and Over 35,000 professionals working in creative industries in-
social issues. (Sturm et al. 2024) and (Boon 2023) critique cluding literature, music, film, theatre and television have
the models of specific music generation platforms, Boomy backed a statement against using unlicensed training data for
and AIVA. (Pasti Da Porto 2023) studies how the music in- AI (Milmo 2024). An open letter from the Artist Rights Al-
dustry can meet UN Sustainable Development Goals. liance advocacy group seeking protections against predatory
use of AI has also been signed by 200 well-known music
Commercial Applications of AI in the Music acts (Artist Rights Alliance 2024).
Industry Meanwhile, a recent survey (Tencer 2024a) of mostly
western music producers suggests that 25% are now using
AI systems are already being used commercially for rec- AI in the creation of music. Of those that do: 74% use it for
ommending music (Born et al. 2021), DJing (Clancy 2021), stem separation; 46% for mastering and EQ plugins; 21% for
separating songs into their separate instrument parts such as generating song elements; and 3% to create entire songs. Of
vocals, drums, guitar etc. (Hennequin et al. 2019; Sun 2023; the 75% not using AI: 82% cite artistic and creative reasons;
Clancy 2021), mastering music (Birtchnell 2018; Robin- 35% quality reasons; 14% costs; and 10% copyright con-
son 2024; Welsh 2022), imitating a singer’s voice (Mon- cerns as reasons. Assistive AI was seen more positively than
roe 2023; Minsker 2021; Coscarelli 2023; Hawthorne 2024), generative AI but both had less than 50% approval. Will-
writing lyrics (Simpson 2022; Taylor 2024) sound design / ingness to pay for AI tools was also low. Another survey of
foley (noa 2021), music transcription (Bittner 2022), cre- 1,600 self-releasing artists (Dalugdug 2023) from DIY dis-
ating concert visuals and VR performances (Google Arts tributor TuneCore found that 27% of indie music artists had
& Culture 2019; Rufo 2024), sample identification (Cetin used AI in some capacity. Of those artists who had used AI
2023), generating music artwork (Jones 2024a), generating tools: 57% had used it to create artwork; 37% had used it
dancer choreography (Studio Wayne McGregor 2019; My- to create promotional assets; and 20% had used it to engage
ers 2023), music venue security and management (Henkin with fans. About half of respondents expressed willingness
2023; Anderson 2024; noa 2019), and music marketing and to license their music for training AI, while a third expressed
public relations (Adgcraft Communications 2024). In the willingness to grant consent for their music, voice or artwork
last two years, there has been a notable rise in the num- to be used in generative AI.
ber and quality of music generation models and services in- Many music industry groups including professional asso-
cluding Suno, Udio, Boomy, AIVA, SOUNDRAW, Tad AI, ciations, music tech companies, music publishers, and aca-
Google’s MusicLM (Agostinelli et al. 2023), Meta’s Au- demic and educational institutions have backed initiatives
dioCraft, Stability AI’s Stable Audio / Open (Evans et al. including aiformusic and the Human Artistry Campaign that
2024a,b) and others. contain principles for AI music creation emphasising re-
Such use cases indicate that AI is already affecting profes- sponsible development and human involvement (Universal
sions in the creative industries including artists, musicians, Music Group 2024; Tencer 2024c). Sony, one of the largest
composers, DJs, visual artists / graphic designers, mixing music publishers in the world with a significant music tech
engineers, marketing, public relations, journalists, sound de- division, has also published a statement declaring its support
signers, song writers, publishers, producers, choreographers, for human artistry and clear intention to opt out of any unli-
and performers. AI systems also influence culture and con- censed AI training or data mining carried out by external ac-
sumer habits (Born et al. 2021; Holzapfel, Sturm, and Co- tors on its content (Aswad 2024). While such organisations
eckelbergh 2018) and exploit general consumers for unpaid are undoubtedly protecting their own commercial interests,
labour (Morreale et al. 2023). We can expect that new appli- they are nevertheless protecting artists’ rights in the process.
General Public / Music Consumers and manufacturing and distributing music equipment (Bren-
Two surveys on the general public’s attitudes to various ap- nan 2020). Some initiatives, guidelines and businesses have
plications of AI in the music industry were conducted by emerged to reduce music’s environmental impact (Nolan
music industry organisations. The first is in the 2023 report 2024; Pasti Da Porto 2023), but widespread industrial adop-
by (International Federation of the Phonographic Industry tion of AI poses challenges to sustainability goals as model
2023) representing the views of over 8000 record companies energy requirements become non-trivial (Peeters 2021).
worldwide. It surveys 43,000 people from 26 countries ac-
counting for 91% of global recorded music market revenues Automation and Deskilling
on their musical habits and opinions including AI. Results Given its significant economic potential, automation in cre-
suggest: 79% feel human creativity essential to the creation ative industries is likely to continue expanding to high-value
of music; 76% think an artist’s music or vocals should not work previously done only by humans. AI systems will not
be used or ingested by AI without permission; 74% agree AI necessarily need to perform better than humans for such
should not be used to clone or impersonate artists without substitution to take place, instead it is likely that quality-
authorisation; 73% agree AI systems should clearly list any cost considerations will inform business adoption (Melville,
music used for training; 70% think there should be restric- Robert, and Xiao 2023). While workers with unique career
tions on what AI can do; and 64% say governments should histories, contextual music industry knowledge and devel-
play a role in setting restrictions in what AI can do. oped ’people skills’ should have the privilege to charge for
The second is a 2024 UK-specific report on AI and the premium services (Birtchnell 2018), certain applications of
Music Industry compiled by UK Music (All-Party Parlia- AI risk displacing and devaluing early career opportunities
mentary Group on Music and UK Music 2024), which in- that are crucial for industry workers to develop confidence,
cluded a poll on the UK general public’s attitudes to mu- experience and portfolio for operating at and obtaining work
sic applications of AI (Whitestone Insight 2024). In it, 83% at a higher level.
agree AI-generated songs should be clearly labelled; 80%
agree that the law should prevent an artist’s music from be- Intellectual Property Issues
ing used to train an AI application without consent; 77% Creative and cultural industries are more than just the gen-
agree that AI-generated music that does not acknowledge eration of IP but IP is of significant interest to various stake-
the original music’s creators amounts to theft; 68% are con- holders due to its economic value (Lee 2022). There are
cerned about music artists losing out financially by having three primary aspects of generative AI applications that in-
their work used by AI to generate new music; 66% are con- tersect with IP protection (Wengen and Ribbert 2024): (i)
cerned about AI generation eventually replacing human cre- learning with protected works as input; (ii) copyright protec-
ativity; 62% are concerned about the rise of deep fakes of tion of AI generated works; and (iii) potential infringement
their favourite artist; and 55% are concerned about listening by the output of pre-existing works. These intersections and
to AI generated music without realising it. their current legal ambiguities are the subject of legal cases
in various countries. For example, major music publishers
Issues Raised by AI in Music UMG, Sony Music, and Warner Records are suing genera-
Experts have argued that AI models will carry benefits such tive AI music companies Suno and Udio in the US for copy-
as increasing reach and accessibility of the arts (Jahromi and right infringement (Brittain 2024), while in the UK a court
Ghazinoory 2023), providing creative opportunities, new case between stock photo provider Getty Images and British
mediums for expression, saving time on routine procedures, AI company Stability AI is pending trial at the High Court
offering inspiration (Deruty et al. 2022; Birtchnell 2018), be- over the use of copyrighted images for training their Sta-
ing a tool for financial benefit, and a way for music fandoms ble Diffusion image generation model (Davies 2024; Brit-
to engage with artists (Shroff 2024). Conversely, issues have tain 2023). Current uncertainty and emerging legislation has
been raised relating to ownership and distribution rights, led AI companies to seek pre-emptive deals to secure usage
royalty sharing, fair use of training data, job displacement of copyrighted materials for training (noa 2024), and ma-
/ automation of traditional creative and knowledge work, jor music publishers such as UMG to seek deals licensing
competition, deskilling, model bias, cultural appropriation, artists’ music and voices with various AI companies includ-
creativity stifling and climate impact (Sturm et al. 2019; ing Google (Ingham 2023), Endel (Olson 2023) and TikTok
Shroff 2024; Rezwana and Maher 2023; Barnett 2023; Hen- (Gerken 2024; Casey 2024). However, the details of these
derson et al. 2020; Boon 2023; Sturm et al. 2024). Exploita- deals mostly remain private.
tive working practices and inequality are already common in Licensing and Remuneration Models Discussions on
the music industry (Arditi and Nolan 2024). Like digitalisa- fair exposure and remuneration structures for music creators
tion (Pusztahelyi and Stefán 2024), AI has the potential to who license their data for training AI models have yet to
both improve existing conditions and pose new issues. generate consensus (Henry et al. 2024). Licensing deals are
often bespoke. For example, Holly Herndon and Grimes are
Environmental Impact two independent artists who have both created and freely
The music industry’s environmental impact is significant distributed their own AI voice models for public use but with
owing to the energy and resource utilisation associated different forms of remuneration. Grimes proposed splitting
with live music, physical and digital music distribution, 50% royalties on any successful song that uses her AI voice
- an identical deal to any human artist she would collaborate hosting user-generated content such as YouTube, Spotify,
with (Monroe 2023) - while income generated from com- or SoundCloud. Content moderation systems on these plat-
mercial licensing of Holly Herndon’s AI voice would go di- forms rely on algorithms that scan millions of content up-
rectly to her IP-owning cooperative to fund new tool devel- loads automatically each day, and can produce outcomes that
opment (Minsker 2021). Furthermore, there are active de- include blocking and taking down material. However, such
bates over opt-in versus opt-out licensing models (Pasquale rigid and widespread enforcement of copyright discourages
and Sun 2024; Gahnberg 2024) - whether data can be used the distribution of infringing works produced through spe-
for AI training by default before user intervention. Critics of cific creative practices that involve sampling, such as hip
opt-out schemes say it is unfair to put the burden of opting hop music, mash-ups, and bootleg remixes (Watson 2024;
out of AI training on the creator whose work is being trained Brøvig-Hanssen and Jones 2023).
on when many will be unaware of such schemes (Milmo Copyright laws that protect owners’ interests in IP are
2024), particularly when a significant proportion of the mu- balanced by limitations and exceptions intended to prevent
sic industry are under-resourced freelancers (Rutter 2016). copyright from excessively impinging on freedom of expres-
Dynamic opt-in licensing and attribution-based models - sion. These are country-specific, for example ‘fair use’ in the
where revenue is paid out proportionally to the use of data US and ’parody’ in the EU / UK, but rarely specify precisely
- have gained traction in industry. For example, music tech what kind of appropriation of copyrighted material is per-
company LANDR’s ‘Fair Trade AI’ program lets musicians mitted. Many uses of copyrighted material are untested in
using its platform earn money by opting their music in to court and their legal status remains unclear. Consequently,
internal AI training (subject to curation) (Robinson 2024). platforms mostly ignore limitations and exceptions to enact
Participating users will receive an attribution-based share of blanket decisions in the interests of operational efficiency,
20% of the revenue generated by any of LANDR’s tools significantly reducing the efficacy of copyright law’s ex-
trained on this dataset. Similarly, Sureel is a dynamic licens- ceptions to the detriment of cultural expression (Brøvig-
ing management platform that tracks data usage for attribu- Hanssen and Jones 2023). Thus, enforcing copyright in these
tion payments and integrates with the Do Not Train registry contexts is a nuanced issue, that requires balancing creative
hosted by Spawning AI, which is respected by Hugging Face liberties with the economic incentives provided by IP rights.
and Stability AI amongst others (Pelczynski 2024). Major
music publishers seem to favour the attribution model, sug- Reproducibility in Music AI Research
gested by the partnership between UMG and the company Advances in music AI research, and AI more generally
ProRata (Stassen 2024; Knibbs 2024a). (Henderson et al. 2018), are currently stifled by repro-
Academics have also proposed alternative models for roy- ducibility issues. New issues have emerged due to the cur-
alty distribution of music created using AI such as (national) rent trend of requiring larger models to achieve current state-
levy-based trust funds and ownership funds (Drott 2021; of-the-art quality and subsequent protection of their com-
Jacques and Flynn 2024), and attribution-based on algorith- mercial value. Producing large commercially-ready (Lavin
mic evaluation (Deng, Zhang, and Ma 2024). et al. 2022) models is only possible with intense resources.
With rare exception, only a few large tech corporations can
Data Poisoning In some cases, bad actors may choose to create and deploy large AI systems at scale, from start to fin-
avoid licensing content or to ignore opt-out directives from ish (Widder, West, and Whittaker 2023). Many companies
creators to train on their data. In response to this, services will also choose to limit access to models or not divulge
have been introduced offering so-called “data poisoning” - proprietary algorithms or model training processes, mostly
imperceptibly altering the pixel composition of images to for commercial reasons but occasionally for AI safety rea-
perturb AI models being trained on that data and degrade sons. For example, the original paper for Google’s Musi-
model performance (Heikkilä 2023; Chen et al. 2023). cLM (Agostinelli et al. 2023) stated that they had “no plans”
Data poisoning can be likened to an adversarial attack that to release the tool to the public due to concerns that a sig-
causes future harm by incorrectly calibrating AI models. It nificantly large proportion of its generated output could be
is challenging to mitigate for (Chen et al. 2023). Though not traced to copyrighted sources. But Google has also recently
yet mature, there is research in applying this to musical au- made it available for beta testing following licensing agree-
dio contexts (Meerza, Sun, and Liu 2024). Without condon- ments with UMG (Jabour 2024). Safety and commercial
ing the practice, it is plain that some of its application stems value are both valid reasons not to share work, but this re-
from creators’ loss of agency over how their work is used. It stricts research progress and conflicts with optimistic hopes
highlights the need for informed consent, fair licensing deals for decentralisation as a benefit of AI (Birtchnell 2018).
over user data, and ethical dataset creation.
Music Identification and IP Enforcement The Google The State of AI Regulation in the West
Assistant software can identify very short music excerpts Optimistic predictions of AI’s economic potential has led
and unearth previously undiscovered samples (Cetin 2023) policy makers around the world to back the technology with
using an approach likely employing neural audio fingerprint- various initiatives (Uren and Edwards 2023). At the same
ing (Arcas et al. 2017). While sample identification can be time, there is caution. G7 Nations have signed the Hiroshima
informative for music listeners, such technology also al- AI Process which contains high-level guiding principles for
lows for easier detection of copyright infringement, espe- developing advanced AI systems (Wintour 2024) and AI leg-
cially when deployed on digital music distribution platforms islation is starting to emerge in many polities including the
EU, US, and UK. While the extent of regulation varies, west- not yet passed. In any case, voluntary measures rarely pro-
ern polities appear motivated in finding a balance between duce a lasting culture of accountability and consistent adop-
supporting their creative / cultural and technology sectors. tion because of reliance on goodwill (Crawford 2024).

European Union The United Kingdom

The EU has recently approved its AI Act, which will be The UK has no specific AI legislation and AI is currently
fully applicable in law by 2026 with some aspects already governed by limited pre-existing laws. For example, the IP
in effect (Official Journal of the European Union 2024). aspect of AI generated works is currently covered by the
General purpose AI models will have to comply with trans- Copyright, Designs and Patents Act 1988. Unlike the US, the
parency requirements and EU copyright law by labelling AI- UK’s act explicitly allows for computer-generated work to
generated content, designing models to prevent generating be copyrighted. Section 9(3) of the act stipulates that when
illegal content, publishing summaries of copyrighted train- a work has ‘no human author’ and is ‘computer-generated’,
ing data, and obtaining the authorisation of the rights holder the ‘author’ is defined as the person who makes ‘arrange-
concerned for any use of copyright protected content, unless ments necessary for the creation of the work’ and is granted
relevant exceptions and limitations apply (Tencer 2024b). copyright. The law currently empowers UK courts to de-
The act was cautiously welcomed by trade organisations cide appropriate authorship of AI-generated music works
for IP rights holders, including a number of music indus- depending on the facts of each case. Ambiguities mean this
try groups (International Federation of the Phonographic In- becomes complex when many different stakeholders could
dustry 2024), but measures were also criticised as “watered be involved in the creation of AI-generated music such as
down” by consumer watchdogs (Corporate Europe Observa- the user, programmer, data controller, training data creator,
tory 2024). The nature of risk self-assessment and disparity model trainer, model owner, investors, or any combination of
in risk-level regulation suggests that many music industry these (Sturm et al. 2019; Koempel 2020; Majumdar 2023).
applications could potentially be assessed as low-risk de- British lawmakers appear to have recognised that the cur-
spite the numerous impacts identified here and consequently rent law is inadequate in the context of AI-generated works.
subjected to looser regulation (Nature Editorials 2024). Ad- In 2023, a debate was held in the UK parliament on IP Rights
ditionally, while the act encourages sustainability through in relation to AI (HC Deb 2023). One of the key outcomes
standardisation, codes of practice and information disclo- was a successful argument against a so-called text and data
sure, it does not respond effectively to the AI industry’s sig- mining (TDM) exemption on copyrighted works to allow AI
nificant environmental impacts (Pereira 2024). developers free use of existing music, literature and works of
art for the purposes of training AI (Culture, Media and Sport
United States of America Committee 2023). However since then the UK Government
has changed following a democratic election in 2024 and it
AI regulation in the US is currently a patchwork of guide- now appears that the TDM exemption and opt-out system are
lines proposed by state and local governments. In terms of being reconsidered by the new government following lobby-
IP, the US Copyright Office released recently updated guide- ing from AI companies (Thomas and Gross 2024).
lines rejecting the notion of considering AI as a contributor,
A report by UK music industry representatives and UK
stating that they do not register works created by machines
parliamentarians (All-Party Parliamentary Group on Music
and creative works still need a human author to qualify for
and UK Music 2024) makes eight recommendations for the
copyright protections (Rockwell 2024; Berkowitz 2024).
UK Government informed by testimony from legal experts
Publicity laws, rather than copyright, protect an individ- on UK, EU, and US IP law, authors associations, and the
ual’s name, image, and likeness from being exploited for British music tech company DAACI. The recommendations
commercial purposes (Rockwell 2024). In 2023 / 2024, the centre around the introduction of a pro-creative industries AI
NO AI FRAUD and NO FAKES Acts were introduced in Bill that protects copyright, introduces new rights and obli-
the US House and Senate respectively seeking to establish gations around labelling and record keeping, and enhances
a “right of publicity” at the federal level and hold individ- personality rights. Other recommendations include transpar-
uals and companies liable for producing or hosting deep- ent labelling requirements for AI-generated content, an obli-
fakes. While these bills have not passed yet, in July 2024 gation for AI developers and those using models to keep a
Tennessee’s ELVIS Act was the first state-level legislation record of training data, compliance with UK copyright law,
to come into effect with the intention to protect musicians addressing the copyright status of AI-generated works, and
having their vocal likeness generated by AI for commer- specific personality rights to protect individual’s voice, im-
cial purposes. The bill makes it illegal to replicate an artist’s age, name, and likeness from misappropriation.
voice without their consent (All-Party Parliamentary Group
on Music and UK Music 2024).
Elsewhere, the AI Environmental Impacts Act has been Conclusions and Recommendations
introduced to direct the National Institute for Standards and AI is affecting the music industries in a myriad of ways still
Technology to collaborate with academia, industry and civil being borne out. This paper has outlined many areas of the
society to establish standards for assessing AI’s environmen- western music industry that are currently affected by AI and
tal impact, and to create a voluntary reporting framework for detailed some of the associated issues, taking into account
AI developers and operators. However, this legislation has current general public and music industry perceptions on
AI and current / proposed legal frameworks in the EU, US recognise that without strong legislation or sufficient incen-
and UK. To conclude this paper, we make recommendations tives many companies and industry workers may prioritise
consistent with the principles of Industry 5.0. on topics of profit acquisition, career opportunities, or user needs over
advocacy, working practices, commercial opportunities, and sustainability goals (Sturm et al. 2024), even when environ-
research directions to support the transition to a sustainable, mentally conscious values are held (Røyseng, Vinge, and
resilient and human-centric music industry. Stavrum 2024). Thus, most importantly, lawmakers should
develop targeted incentives in relation to AI and the music
Sustainability and Ethical Practice industry to address the climate crisis (Crawford 2024).
Disclosure of potential impacts and ethical implications of
AI in music is currently lacking. Findings in (Barnett 2023) Investment in Human Creativity and Industry
suggest that less than 10% of generative audio research pa- Grassroots
pers discuss any negative broader impact in their work, even Large, data-intensive AI models have shifted the value of
though 65% consider potential positive broader impacts. music towards its profitability as data, rather than intrinsic
(Henderson et al. 2020) meanwhile found substantial under- artistic worth. But it is crucial to acknowledge that the effi-
reporting of information to calculate energy and resource cacy of these AI systems is contingent upon the high-quality
use. (Reje 2022) suggests that many emerging music tech materials from which they learn that predominantly stem
start-ups developing AI do not prioritise the adoption of for- from human creativity (Jacques and Flynn 2024). There-
mal ethical guidelines and (Oğul 2024) notes that environ- fore, we recommend that individuals and organisations that
mental impact and sustainability are frequently missing even feel strongly about work quality and Industry 5.0. princi-
from published ethical guidelines on AI in the music indus- ples should strive to commission human creatives where
try. Largely, AI developers are not prioritising energy reduc- practicable. Similarly, we would like to see more initiatives
tion, energy-efficient models, or disclosing relevant data. As from industry and governments investing in the music in-
such it is hard to get accurate and complete data on AI’s en- dustry grassroots to maintain a creative talent pipeline. For
vironmental impact (Crawford 2024). example, Spotify’s investment in UK youth clubs (Collins
Organisations developing AI music systems should be 2024) and the UK Government’s backing of a voluntary levy
transparent and share relevant information and resources on tickets at large venues to support grassroots venues and
where possible, disclosing potential impacts in appropriate workers (Reilly 2024). Such initiatives will not only main-
language (Haueis 2024). They should create or adopt ethical tain fulfilling employment opportunities, facilitate worker
guidelines (e.g. aiformusic) to guide development and fol- upskilling, and build music industry resilience, but also con-
low the machine learning technology readiness level (ML- tribute to the resilience of AI models in general. New, high
TRL) assessment framework proposed by (Lavin et al. 2022) quality dataset creation will help to maintain model per-
which prioritises ethics and fairness to develop principled, formance and contemporaneity while web-scraped data be-
safe, and trusted AI technology. Model developers should comes increasingly unreliable and ineffective due to AI-
take steps to report the energy required for the training and generated content (Alemohammad et al. 2024; Jones 2024b).
inference of their AI models and make efforts to minimise
energy use through adoption of efficient model architectures Intellectual Property Licensing and Compliance
and data pipelines (Chen et al. 2023; Douwes et al. 2023), Shifting legal situations in the UK, EU, and US in favour of
following examples such as (Douwes et al. 2023; Douwes the entrenched music industries demonstrate the importance
and Serizel 2024; Tabata and Wang 2021; Utz and DiPaola for individuals and organisations to organise, advocate, and
2023) which demonstrate methods for computing and op- lobby for their position. However, developments of both the
timising energy use and emissions in the training and in- technology and the legal frameworks governing it should be
ference of AI audio models and other digital audio applica- closely watched to ensure they are fair to creators and other
tions. Developers should also consider using smaller training key industry workers. Standard licensing practices are yet to
datasets which require less energy to train on and enable ad- emerge so it is important to set fair precedents within the in-
ditional benefits in certain use-cases (Vigliensoni, Perry, and dustry in terms of remuneration, transparency, and consent.
Fiebrink 2022). These measures will enable adopters to, in A global music industry with divergent legislation makes
turn, make an informed decision about specific model use. international data sharing and standards adherence com-
We agree with (Brennan 2020) and (Crawford 2024) that plex. There are opportunities in IP licensing and environ-
addressing the environmental impact of AI is a collective ef- mental standards compliance certification and enforcement
fort from industry, researchers, legislators, and the public. as demonstrated by the Fairly Trained Initiative (Knibbs
Audiences must consider their consumption choices. Like- 2024b). Several ongoing research directions could be valu-
wise musicians, manufacturers, promoters, labels, and tech- able to various aspects of compliance from detecting deep
nology companies that rely on AI or musical content for fakes and AI-generated music (Vaglio et al. 2021; Khanjani,
their business model should consider the sustainability of Watson, and Janeja 2023; Desblancs et al. 2024), to assess-
their working practices and shift towards more sustainable ing generative music output similarity for attribution-based
ones where practicable. For this, we recommend leverag- models (Flexer 2023; Batlle-Roca et al. 2024; Deng, Zhang,
ing resources from organisations such as Music Declares and Ma 2024), the placement and detection of watermarks in
Emergency (Nolan 2024) and elsewhere (Jones, McLachlan, AI-generated content (Porter 2023), and data poisoning for
and Mander 2021) to inspire mitigation efforts. However, we music (Meerza, Sun, and Liu 2024).
 Acknowledgements Batlle-Roca, R.; Liao, W.-H.; Serra, X.; Mitsufuji, Y.; and
Alexander Williams is a research student at the UKRI Cen- Gómez, E. 2024. Towards Assessing Data Replication in
tre for Doctoral Training in Artificial Intelligence and Mu- Music Generation with Music Similarity Metrics on Raw
sic, supported jointly by UK Research and Innovation [grant Audio. In Proc. of the 25th Int. Society for Music Infor-
number EP/S022694/1] and Queen Mary University of Lon- mation Retrieval Conf. San Francisco, CA, USA.
don. We wish to thank Lord Tim Clement-Jones for an in- Berkowitz, A. E. 2024. ”Gimme Some Truth”: AI Music
sightful discussion on AI governance in the creative indus- and Implications for Copyright and Cataloging. Information
tries. The Centre for Digital Music is a signatory to the aifor- Technology and Libraries, 43(3).
music initiative and member of Music Technology UK. Birtchnell, T. 2018. Listening Without Ears: Artificial In-
telligence in Audio Mastering. Big Data & Society, 5(2):
References 205395171880855.
2019. Digital Barriers enhances security operations at Lon- Bittner, R. 2022. Meet Basic Pitch: Spotify’s Open Source
don’s O2 Arena for The BRIT Awards and National Televi- Audio-to-MIDI Converter. Spotify Engineering.
sion Awards. Source Security. Boon, H. 2023. Alien power chords: AIVA has ‘musical
2021. Queen Mary spinout Nemisindo launches online artist status’ in France – but what about the humans who
sound design service based on procedural audio technology. feed it? The Sociological Review Magazine.
Queen Mary University of London. Born, G.; Morris, J.; Diaz, F.; and Anderson, A. 2021. Artifi-
cial Intelligence, Music Recommendation, and the Curation
2024. OpenAI signs multi-year content partnership with
of Culture: A White Paper.
Condé Nast. The Guardian.
Brennan, M. 2020. The Environmental Sustainability of
Adgcraft Communications. 2024. Integrating AI in Music the Music Industries. In Oakley, K.; and Banks, M., eds.,
Industry PR. Cultural Industries and the Environmental Crisis: New Ap-
Agostinelli, A.; Denk, T. I.; Borsos, Z.; Engel, J.; Verzetti, proaches for Policy, 37–49. Cham: Springer International
M.; Caillon, A.; Huang, Q.; Jansen, A.; Roberts, A.; Publishing. ISBN 978-3-030-49384-4.
Tagliasacchi, M.; Sharifi, M.; Zeghidour, N.; and Frank, Brittain, B. 2023. Getty Images lawsuit says Stability AI
C. 2023. MusicLM: Generating Music From Text. misused photos to train AI. Reuters.
ArXiv:2301.11325 [cs, eess]. Brittain, B. 2024. Music labels’ AI lawsuits create copyright
Alemohammad, S.; Casco-Rodriguez, J.; Luzi, L.; Imtiaz, puzzle for courts. Reuters.
A.; Babaei, H.; LeJeune, D.; Siahkoohi, A.; and Baraniuk, Brøvig-Hanssen, R.; and Jones, E. 2023. Remix’s retreat?
R. 2024. Self-Consuming Generative Models go MAD. In Content moderation, copyright law and mashup music. New
The Twelfth International Conference on Learning Repre- Media & Society, 25(6): 1271–1289. SAGE Publications.
sentations (ICLR). Vienna, Austria.
Casey, S. 2024. Tik Tok and Universal Music Group’s Li-
All-Party Parliamentary Group on Music; and UK Music. censing Battle. Cardozo Arts & Entertainment Law Journal
2024. Artificial Intelligence and the Music Industry – Master Blog.
or Servant? Technical report, United Kingdom. Cetin, M. 2023. Detecting samples less than one second long
Anderson, C. 2024. The O2 introduces first self-serve bars now possible with Google Assistant, report shows. DJ Mag.
powered by AI. News Shopper. Chen, Z.; Wu, M.; Chan, A.; Li, X.; and Ong, Y.-S. 2023.
Arcas, B. A. y.; Gfeller, B.; Guo, R.; Kilgour, K.; Kumar, S.; Survey on AI Sustainability: Emerging Trends on Learn-
Lyon, J.; Odell, J.; Ritter, M.; Roblek, D.; Sharifi, M.; and ing Algorithms and Research Challenges [Review Article].
Velimirović, M. 2017. Now Playing: Continuous low-power IEEE Computational Intelligence Magazine, 18(2): 60–77.
music recognition. In Machine Learning on the Phone Work- Cisac; and PMP Strategy. 2024. Study on the Economic
shop at the 31st Conference on Neural Information Process- Impact of Generative AI in the Music and Audiovisual In-
ing Systems. Long Beach, CA, USA. ArXiv:1711.10958. dustries. Technical report.
Arditi, D.; and Nolan, R., eds. 2024. The Palgrave Hand- Clancy, M. 2021. Reflections on the Financial and Ethical
book of Critical Music Industry Studies. Cham: Springer Implications of Music Generated by Artificial Intelligence.
Nature Switzerland. ISBN 978-3-031-64012-4 978-3-031- PhD Thesis, Trinity College, Dublin, Ireland.
64013-1. Collins, R. 2024. Can Spotify’s youth club plan actually help
Artist Rights Alliance. 2024. 200+ Artists Urge Tech Plat- the future of music? BBC News.
forms: Stop Devaluing Music. Corporate Europe Observatory. 2024. Trojan Horses: How
Aswad, J. 2024. Sony Music Warns AI Developers Not to European Startups Teamed up with Big Tech to Gut the AI
Use Its Content for Training. Variety. Act.
Barnett, J. 2023. The Ethical Implications of Generative Au- Coscarelli, J. 2023. An A.I. Hit of Fake ‘Drake’ and ‘The
dio Models: A Systematic Literature Review. In Proceed- Weeknd’ Rattles the Music World. The New York Times.
ings of the 2023 AAAI/ACM Conference on AI, Ethics, and Crawford, K. 2024. Generative AI’s environmental costs are
Society, AIES ’23, 146–161. Montréal, QC, Canada.: Asso- soaring — and mostly secret. Nature, 626(8000): 693–693.
ciation for Computing Machinery. ISBN 9798400702310. Nature Publishing Group.
Crompton, L. 2021. The decision-point-dilemma: Yet an- Google Arts & Culture. 2019. Editorial Feature | How an
other problem of responsibility in human-AI interaction. Artist Used AI to Make a Concert Hall Dream.
Journal of Responsible Technology, 7-8: 100013. Haueis, P. 2024. Climate concepts for supporting political
Culture, Media and Sport Committee. 2023. Connected goals of mitigation and adaptation: The case for “climate
tech: AI and creative technology. Technical Report HC crisis”. WIREs Climate Change, 15(5): e893.
1643, House of Commons, United Kingdom. Hawthorne, K. 2024. ‘I’m empowering my song to go and
Dalugdug, M. 2023. 27% of indie artists have used AI music make love with different people’: Imogen Heap on how her
tools, according to TuneCore study. Music Business World- AI twin will rewrite pop. The Guardian.
wide. HC Deb. 2023. Artificial Intelligence: Intellectual Property
Davies, C. W. 2024. Getty Images V Stability I: The Im- Rights - Hansard - UK Parliament.
plications for UK Copyright Law and Licensing. Pinsent Heikkilä, M. 2023. This new data poisoning tool lets artists
Masons. fight back against generative AI. MIT Technology Review.
Deng, J.; Zhang, S.; and Ma, J. 2024. Computational Copy- Henderson, P.; Hu, J.; Romoff, J.; Brunskill, E.; Jurafsky,
right: Towards A Royalty Model for Music Generative AI. D.; and Pineau, J. 2020. Towards the systematic reporting of
In 2nd Workshop on Generative AI and Law, co located with the energy and carbon footprints of machine learning. The
the International Conference on Machine Learning. Hon- Journal of Machine Learning Research, 21(1): 248:10039–
olulu, Hawaii, USA. 248:10081.
Department for Education. 2023. The Impact of AI on UK Henderson, P.; Islam, R.; Bachman, P.; Pineau, J.; Precup,
Jobs and Training. Technical report. D.; and Meger, D. 2018. Deep reinforcement learning that
Deruty, E.; Grachten, M.; Lattner, S.; Nistal, J.; and matters. In Proceedings of the Thirty-Second AAAI Confer-
Aouameur, C. 2022. On the Development and Practice of ence on Artificial Intelligence and Thirtieth Innovative Ap-
AI Technology for Contemporary Popular Music Produc- plications of Artificial Intelligence Conference and Eighth
tion. Transactions of the International Society for Music AAAI Symposium on Educational Advances in Artificial In-
Information Retrieval, 5(1): 35–49. Ubiquity Press. telligence, AAAI’18/IAAI’18/EAAI’18, 3207–3214. New
Desblancs, D.; Meseguer-Brocal, G.; Hennequin, R.; and Orleans, Louisiana, USA: AAAI Press. ISBN 978-1-57735-
Moussallam, M. 2024. From Real to Cloned Singer Iden- 800-8.
tification. In Proc. of the 25th Int. Society for Music In- Henkin, D. 2023. Orchestrating The Future—AI In The Mu-
formation Retrieval Conf. San Francisco, CA, USA: arXiv. sic Industry. Forbes.
ArXiv:2407.08647. Hennequin, R.; Khlif, A.; Voituret, F.; and Moussallam, M.
Douwes, C.; Bindi, G.; Caillon, A.; Esling, P.; and Briot, J.- 2019. Spleeter: A Fast and State-of-the Art Music Source
P. 2023. Is Quality Enough? Integrating Energy Consump- Separation Tool with Pre-Trained Models. In Late Break-
tion in a Large-Scale Evaluation of Neural Audio Synthesis ing/Demo at the 20th International Society for Music Infor-
Models. In ICASSP 2023 - 2023 IEEE International Confer- mation Retrieval. Delft, The Netherlands.
ence on Acoustics, Speech and Signal Processing (ICASSP), Henry, A.; Wiratama, V.; Afilipoaie, A.; Ranaivoson, H.; and
1–5. ISSN: 2379-190X. Arrivé, E. 2024. Impacts of AI on Music Consumption and
Douwes, C.; and Serizel, R. 2024. From Computation Fairness. Emerging Media, 27523543241269047. SAGE
to Consumption: Exploring the Compute-Energy Link for Publications.
Training and Testing Neural Networks for SED Systems. In Holzapfel, A.; Sturm, B. L.; and Coeckelbergh, M. 2018.
Proceedings of the 9th Workshop on Detection and Classifi- Ethical Dimensions of Music Information Retrieval Tech-
cation of Acoustic Scenes and Events. Tokyo, Japan. nology. Transactions of the International Society for Music
Drott, E. 2021. Copyright, compensation, and commons in Information Retrieval, 1(1): 44–55.
the music AI industry. Creative Industries Journal, 14(2): Hopster, J. 2021. What are socially disruptive technologies?
190–207. Routledge. Technology in Society, 67: 101750.
Evans, Z.; Carr, C. J.; Taylor, J.; Hawley, S. H.; and Pons, Horowitz, M. C.; Kahn, L.; Macdonald, J.; and Schneider,
J. 2024a. Fast Timing-Conditioned Latent Audio Diffusion. J. 2024. Adopting AI: how familiarity breeds both trust and
ArXiv:2402.04825. contempt. AI & SOCIETY, 39(4): 1721–1735.
Evans, Z.; Parker, J. D.; Carr, C. J.; Zukowski, Z.; Taylor, J.; Huang, R.; Sturm, B. L. T.; and Holzapfel, A. 2021. De-
and Pons, J. 2024b. Stable Audio Open. ArXiv:2407.14358. Centering the West: East Asian Philosophies and the Ethics
Flexer, A. 2023. Can ChatGPT Be Useful for Distant Read- of Applying Artificial Intelligence to Music. In Proc. of the
ing of Music Similarity? In 2nd Workshop on Human- 22nd Int. Society for Music Information Retrieval Conf. On-
Centric Music Information Research. Milan, Italy. line.
Gahnberg, C. 2024. AI-Control: Opt-Out Mechanisms From Huang, R. S.; Holzapfel, A.; Sturm, B. L. T.; and Kaila, A.-
the View of a Governance Cycle. In IAB Workshop on AI- K. 2023. Beyond Diverse Datasets: Responsible MIR, In-
CONTROL. terdisciplinarity, and the Fractured Worlds of Music. Trans-
Gerken, T. 2024. TikTok and Universal settle dispute over actions of the International Society for Music Information
music royalties. BBC News. Retrieval, 6(1): 43–59.
Ingham, T. 2023. YouTube and Universal Music Group part- R.; Wu, S.; Wu, S.-L.; Dai, S.; Lei, S.; Kang, S.; Dixon, S.;
ner to develop AI music tools – complete with ‘protections’ Chen, W.; Huang, W.; Du, X.; Qu, X.; Tan, X.; Li, Y.; Tian,
for artists and rightsholders. Music Business Worldwide. Z.; Wu, Z.; Wu, Z.; Ma, Z.; and Wang, Z. 2024. Foundation
International Federation of the Phonographic Industry. 2023. Models for Music: A Survey. ArXiv:2408.14340.
Engaging With Music 2023. Technical report. Majumdar, A. 2023. Facing the Music: The Future of Copy-
International Federation of the Phonographic Industry. 2024. right Law and Artificial Intelligence in Music Industry.
EU AI ACT - Joint statement from European creators and Malloch, S.; and Trevarthen, C. 2018. The Human Nature of
rightsholders. Music. Frontiers in Psychology, 9. Frontiers.
Jabour, G. 2024. Drake Or Fake? Perceptions, Concerns, Mbamba, U. O. L. 2024. Impact of Selected Fourth Indus-
and Business Implications of AI-Generated Vocals. Ph.D. trial Revolution Technologies on the Music Industry: An Ex-
thesis, University of Texas, Austin, TX, USA. ploration of the Pros and Cons. Umma Journal of Contem-
Jacques, S.; and Flynn, M. 2024. Protecting Human Cre- porary Literature and Creative Arts, 11(1): 179–209.
ativity in AI-Generated Music with the Introduction of an Meerza, S. I. A.; Sun, L.; and Liu, J. 2024. HARMONY-
AI-Royalty Fund. GRUR International, 73(12): 1137–1149. CLOAK: Making Music Unlearnable for Generative AI. 85–
Jahromi, G. S.; and Ghazinoory, S. 2023. How to use bits 85. IEEE Computer Society. ISBN 9798331522360. ISSN:
for beats: the future strategies of music companies for using 2375-1207.
Industry 4.0 technologies in their value chain. Information Melville, N. P.; Robert, L.; and Xiao, X. 2023. Putting hu-
Systems and e-Business Management, 21(3): 505–525. mans back in the loop: An affordance conceptualization of
Jones, A. 2024a. Tears For Fears Address Use Of AI For the 4th industrial revolution. Information Systems Journal,
New Album Cover. Stereogum. 33(4): 733–757.
Jones, C.; McLachlan, C.; and Mander, S. 2021. Super-Low Milmo, D. 2024. Thom Yorke and Julianne Moore join thou-
Carbon Live Music: a roadmap for the UK live music sec- sands of creatives in AI warning. The Guardian.
tor to play its part in tackling the climate crisis. Technical Minsker, E. 2021. Holly Herndon’s AI Deepfake “Twin”
Report SLCM 5.1, Tyndall Centre for Climate Change Re- Holly+ Transforms Any Song Into a Holly Herndon Song.
search, United Kingdom. Pitchfork.
Jones, N. 2024b. The AI revolution is running out of data. Monroe, J. 2023. Grimes Unveils Software to Mimic Her
What can researchers do? Nature, 636(8042): 290–292. Na- Voice, Offering 50-50 Royalties for Commercial Use. Pitch-
ture Publishing Group. fork.
Khanjani, Z.; Watson, G.; and Janeja, V. P. 2023. Audio Morreale, F.; Bahmanteymouri, E.; Burmester, B.; Chen, A.;
deepfakes: A survey. Frontiers in Big Data, 5. Frontiers. and Thorp, M. 2023. The unwitting labourer: extracting hu-
Knibbs, K. 2024a. Generative AI Has a ’Shoplifting’ Prob- manness in AI training. AI & SOCIETY. Springer.
lem. This Startup CEO Has a Plan to Fix It. Wired. Myers, A. 2023. AI-powered EDGE Dance Animator Ap-
Knibbs, K. 2024b. This Tech Exec Quit His Job to Fight plies Generative AI to Choreography.
Generative AI’s Original Sin. Wired. Nature Editorials. 2024. There are holes in Europe’s AI Act
Koempel, F. 2020. From the gut? Questions on Artificial — and researchers can help to fill them. Nature, 625(7994):
Intelligence and music. Queen Mary Journal of Intellectual 216–216.
Property, 10(4): 503–513. Nolan, R. 2024. Music Declares an Emergency: Music In-
Lavin, A.; Gilligan-Lee, C. M.; Visnjic, A.; Ganju, S.; New- dustry Studies in the Context of a Changing Climate. In
man, D.; Ganguly, S.; Lange, D.; Baydin, A. G.; Sharma, A.; Arditi, D.; and Nolan, R., eds., The Palgrave Handbook of
Gibson, A.; Zheng, S.; Xing, E. P.; Mattmann, C.; Parr, J.; Critical Music Industry Studies, 525–535. Cham: Springer
and Gal, Y. 2022. Technology readiness levels for machine Nature Switzerland. ISBN 978-3-031-64013-1.
learning systems. Nature Communications, 13(1): 6039. Na- Official Journal of the European Union. 2024. Regula-
ture Publishing Group. tion (EU) 2024/1689 of the European Parliament and of the
Lee, H.-K. 2022. Rethinking creativity: creative industries, Council of 13 June 2024 Laying down Harmonised Rules
AI and everyday creativity. Media, Culture & Society, 44(3): on Artificial Intelligence and Amending Regulations (EC)
601–612. SAGE Publications Ltd. No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU)
Lerch, A. 2018. The Relation Between Music Technol- 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Direc-
ogy and Music Industry. In Bader, R., ed., Springer Hand- tives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Ar-
book of Systematic Musicology, 899–909. Berlin, Heidel- tificial Intelligence Act) (Text with EEA Relevance). Leg-
berg: Springer. ISBN 978-3-662-55004-5. islative Body: CONSIL, EP.
Ma, Y.; Øland, A.; Ragni, A.; Sette, B. M. D.; Saitis, C.; Olson, C. A. 2023. World’s Largest Music Company Em-
Donahue, C.; Lin, C.; Plachouras, C.; Benetos, E.; Shatri, braces AI. Forbes.
E.; Morreale, F.; Zhang, G.; Fazekas, G.; Xia, G.; Zhang, H.; Oğul, S. 2024. In Tune with Ethics: Responsible Artificial
Manco, I.; Huang, J.; Guinot, J.; Lin, L.; Marinelli, L.; Lam, Intelligence and Music Industry. REFLEKTİF Sosyal Bilim-
M. W. Y.; Sharma, M.; Kong, Q.; Dannenberg, R. B.; Yuan, ler Dergisi, 5(1): 139–149.
Pasquale, F.; and Sun, H. 2024. Consent and Compensation: Simpson, D. 2022. ’It feels like a fresh start’: why Every-
Resolving Generative AI’s Copyright Crisis. Virginia Law thing Everything turned to AI to write their new album. The
Review, 110. Guardian.
Pasti Da Porto, V. C. 2023. Sustainability and Inclusion: Stassen, M. 2024. Universal Music strikes strategic agree-
How SDGs may be implemented in the Music Industry. ment with AI startup ProRata, which just raised $25m for a
MA Thesis in Management, Universitá Ca’Foscari Venezia, chatbot and tech to attribute and compensate content owners.
Italy. Accepted: 2023-02-17 Publisher: Università Ca’ Fos- Music Business Worldwide.
cari Venezia. Studio Wayne McGregor. 2019. Living Archive: Creating
Peeters, G. 2021. The Deep Learning Revolution in MIR: Choreography with Artificial Intelligence.
The Pros and Cons, the Needs and the Challenges. In Sturm, B. L. T.; Déguernel, K.; Huang, R. S.; Kaila, A.-K.;
Kronland-Martinet, R.; Ystad, S.; and Aramaki, M., eds., Jääskeläinen, P.; Kanhov, E.; Cros Vila, L.; Dalmazzo, D.;
Perception, Representations, Image, Sound, Music, Lecture Casini, L.; Bown, O. R.; Collins, N.; Drott, E.; Sterne, J.;
Notes in Computer Science, 3–30. Cham: Springer Interna- Holzapfel, A.; and Ben-Tal, O. 2024. AI Music Studies:
tional Publishing. ISBN 978-3-030-70210-6. Preparing for the Coming Flood. In AI Music Conference.
Pelczynski, M. M. 2024. The AI Era: Building Sustain- Oxford, United Kingdom.
able AI Business Models for the Music Industry. Forms + Sturm, B. L. T.; Iglesias, M.; Ben-Tal, O.; Miron, M.; and
Shapes. Gómez, E. 2019. Artificial Intelligence and Music: Open
Questions of Copyright Law and Engineering Praxis. Arts,
Pereira, J. R. L. d. 2024. The EU AI Act and environmental 8(3): 115. Multidisciplinary Digital Publishing Institute.
protection: the case for a missed opportunity. Heinrich-Böll-
Stiftung. Suchman, L. 2023. The uncontroversial ‘thingness’ of AI.
Big Data & Society, 10(2): 20539517231206794. SAGE
Porter, J. 2023. Google is embedding inaudible watermarks Publications Ltd.
right into its AI generated music. The Verge.
Sun, M. 2023. Paul McCartney says there’s nothing artificial
Pusztahelyi, R.; and Stefán, I. 2024. Improving Industry 4.0 in new Beatles song made using AI. The Guardian.
to Human-Centric Industry 5.0 in Light of the Protection of Tabata, T.; and Wang, T. Y. 2021. Life Cycle Assessment
Human Rights. In 25th International Carpathian Control of CO2 Emissions of Online Music and Videos Streaming
Conference (ICCC), 1–6. in Japan. Applied Sciences, 11(9): 3992. Multidisciplinary
Reilly, N. 2024. The UK Government Has Officially Backed Digital Publishing Institute.
a Levy on Stadium and Arena Tickets. Rolling Stone UK. Taylor, J. 2024. Suno AI can generate power ballads about
Reje, A. 2022. Ethical Risk Analysis of the Use of AI in Mu- coffee – and jingles for the Guardian. But will it hurt musi-
sic Production. Ph.D. thesis, KTH Royal Institute of Tech- cians? The Guardian.
nology, Stockholm, Sweden. Tencer, D. 2024a. 25% of music producers are now using
Rezwana, J.; and Maher, M. L. 2023. User Perspectives AI, survey says – but a majority shows strong resistance.
on Ethical Challenges in Human-AI Co-Creativity: A De- Music Business Worldwide.
sign Fiction Study. In Proceedings of the 15th Conference Tencer, D. 2024b. As landmark AI Act passes EU parlia-
on Creativity and Cognition, C&C ’23, 62–74. New York, ment vote, rightsholders urge ‘meaningful and effective’ en-
NY, USA: Association for Computing Machinery. ISBN forcement of copyright. Music Business Worldwide.
9798400701801. Tencer, D. 2024c. Music industry applauds introduction
Robinson, K. 2024. LANDR’s ‘Fair Trade AI’ Program Lets of ‘No AI FRAUD Act’ in US Congress. Music Business
Musicians Earn Money by Contributing to AI Training. Bill- Worldwide.
board. Thomas, D.; and Gross, A. 2024. UK to consult on ‘opt-
Rockwell, E. 2024. The Heart of Artificial Intelligence in the out’ model for AI content-scraping in blow to publishers.
Music Industry: Amending the Music Modernization Act to Financial Times.
Promote Transparency. Boston College Intellectual Property UK Music. 2024. This Is Music 2024. Technical report,
and Technology Forum, 2024. United Kingdom.
Rufo, Y. 2024. Elvis Evolution: Presley to be brought to life Universal Music Group. 2024. Music Industry Unites to Pro-
using AI for new immersive show. BBC News. tect the Rights of Musicians Amid the Growth of Generative
AI Technology.
Rutter, P. 2016. The Music Industry Handbook. Routledge,
Uren, V.; and Edwards, J. S. 2023. Technology readiness
2nd edition.
and the organizational journey towards AI adoption: An em-
Røyseng, S.; Vinge, J.; and Stavrum, H. 2024. The cultural pirical study. International Journal of Information Manage-
dissonance of sustainable live music. Annals of Leisure Re- ment, 68: 102588.
search, 1–15. Routledge. Utz, V.; and DiPaola, S. 2023. Climate Implications of
Shroff, L. 2024. AI & Copyright: A Case Study of the Music Diffusion-based Generative Visual AI Systems and their
Industry. GRACE: Global Review of AI Community Ethics, Mass Adoption. In 14th International Conference on Com-
2(1). putational Creativity. Ontario, Canada.
Vaglio, A.; Hennequin, R.; Moussallam, M.; and Richard, G.
2021. The Words Remain the Same: Cover Detection with
Lyrics Transcription. In Proc. of the 22nd Int. Society for
Music Information Retrieval Conf. Online.
Vigliensoni, G.; Perry, P.; and Fiebrink, R. 2022. A Small-
Data Mindset for Generative AI Creative Work. In Genera-
tive AI and HCI Workshop at CHI 2022.
Watson, A.; and Leyshon, A. 2022. Negotiating platformisa-
tion: MusicTech, intellectual property rights and third wave
platform reintermediation in the music industry. Journal of
Cultural Economy, 15(3): 326–343. Routledge.
Watson, J. 2024. Copyright and the Production of Hip Hop
Music.
Welsh, A. C. 2022. iZotope announces Ozone 10 AI master-
ing suite. DJMag.
Wengen, W. v.; and Ribbert, R. 2024. EU AI Acts Opt-Out
Trend May Limit Data Use for Training AI Models. Green-
berg Traurig.
Whitestone Insight. 2024. UK Music AI Poll – 27th March
2024.
Widder, D. G.; West, S.; and Whittaker, M. 2023. Open (For
Business): Big Tech, Concentrated Power, and the Political
Economy of Open AI.
Wintour, P. 2024. Why the pope has the ears of G7 leaders
on the ethics of AI. The Guardian.
Xu, X.; Lu, Y.; Vogel-Heuser, B.; and Wang, L. 2021. In-
dustry 4.0 and Industry 5.0—Inception, conception and per-
ception. Journal of Manufacturing Systems, 61: 530–535.