Keeping Ownership Authentic: AI and Control in the Modern Matrix of Music

Emma Hewins looks at the questions of creativity, authenticity, control and how creators can assert ownership as AI generated music takes hold

“Music was my refuge. I could crawl into the space between
the notes and curl my back to loneliness.”
Maya Angelou

Maya Angelou’s words still explain why music matters: it is more than entertainment, but a source of belonging, escape and identity. For example in The Phantom of the Opera, music becomes an outsider’s defining refuge, shaped by sound and architecture. Today, that refuge is becoming more digital, especially as AI-generated songs span streaming platforms and social media, making it harder to tell what is human-made, authorised, and properly owned.

This is not only a creative problem. It is a commercial and legal one. With global entertainment and media revenues projected to reach US$3.5 trillion by 2029^[1], the stakes are significant. Some artists have embraced AI’s potential: Google DeepMind’s Dream Track includes artists such as Charli XCX, who consented to help develop its model, Lyria. Similarly, Verses, created by Sean Lee and Kyungtae Kim, points to a hybrid future of human input and interactive AI audio output. Additionally, RaagaPay aims to build ethical, royalty-based datasets for Hindustani classical music. Others are more wary. Thomas Bangalter of Daft Punk has emphasised emotional authenticity over machine output, while virtual artist Mya Blue has highlighted risks of cultural appropriation, unequal profit-sharing and harm to traditional African music. Beneath all this, the real question is no longer just what we hear, but how we distinguish reality from simulation in the first place.

In music, however, that distinction quickly becomes operational: it concerns who trained the model, whose works were used, who approved the output, how is it labelled, and who gets paid. This article looks at where we are in the debate over authenticating human versus AI content, and at the emerging legal and technical infrastructure that might make authenticity enforceable. It covers: (1) consent and control in AI training and outputs; (2) licensing and data systems that allow rights to travel; and (3) provenance tools, including NFT-style systems, that can record and verify authorised origin and usage.

The Importance of Authenticity: Music as IP
AI in music means the legal issue is no longer just who owns a track, but how it was trained, created, licensed and controlled. Authenticity endures only if artists retain meaningful control and ownership can be verified. Critics worry that immersion, simulation and AI-generated performance may weaken music’s human connection. However, musicologist Matthew Lavy for instance suggests listeners hear music as a kind of narrative, integrating sound, utterance and context into a coherent structure^[2]. In connection, CBS Songs Ltd v Amstrad Consumer Electronics Plc showed that supplying copying technology does not, without more, amount to authorising infringement. Similarly authenticity is not opposed to technology; it depends on whether technology preserves a credible link between work, creator and audience.

The production architecture behind modern music turns that feeling into an IP issue. Digitally, composers can use all sorts of software from scorewriters like Sibelius, digital audio workstations such as Ableton Live or Logic Pro, or DJ software such as rekordbox to create their work. Dolby Atmos serves as the industry standard for sound by letting it move as objects within a 3D field. The key issue in all of this is not ownership alone but whether human contribution, authorisation, and rights remain identifiable.

Legally in the UK, under the Copyright, Designs and Patents Act 1988 (CDPA), rights may subsist in the musical work (s.3) and sound recording (s.5A), and exploitation may engage restricted acts (s.16), including copying (s.17) and adaptation (s.21)^[3]. In practice, however, the commercial focus is on who holds the chain of title, approvals, credits, and royalty mechanics for each layer, including stems, masters, session files, and platform-specific builds. This makes AI music licensing especially important, because commercial use depends on aligned rights in source works, outputs, and downstream uses. The UK’s Collective Management of Copyright (EU Directive) Regulations 2016 support that framework by requiring collective management organisations (CMOs) to act in right holders’ interests and impose only necessary obligations (reg 3), allowing right holders to choose rights managers and retain control (reg 4), and facilitating multi-territorial licensing through interoperable data systems (regs 23–27, 30)^[4]. Without that infrastructure, permissions, royalties, and performances may be lost or unprotected.

AI and Consent
AI is now the central pressure point. After years of copyright disputes over sampling and similarity, the industry is facing a larger question: whether machine-generated music can be built on borrowed human work without consent. Reports of artists having AI-generated tracks uploaded under their name without permission, show how easily authorship can be diluted and revenues diverted, especially for independent artists with limited enforcement resources. At the same time, tools such as Suno AI, which can generate full tracks from simple prompts, have intensified concerns about how quickly synthetic music can scale. YouTube duo TwoSet Violin found Suno effective for pop music output whilst finding classical music AI tools like NotaGen precise yet predictable. In 2024, Warner Records, Sony Music and Universal Music Group sued Suno and Udio for large-scale copyright infringement, while GEMA also pursued action against Suno in Europe.

There are, however, more consent-forward models. Stanford PhD graduate Holly Herndon offers one of the clearest examples: Spawn (her AI model on the album Proto) was trained on authorised choir recordings; Holly+ allows her voice to be used by others within a controlled framework; and HaveIBeenTrained helps creators check whether their work has been used without consent. Herndon also uses a decentralised autonomous organisation (DAO) in which token holders vote on model use and share revenues while she retains creative control. The point is not that AI becomes risk-free but that authentic AI in music depends on governance that can be evidenced.

Within UKIPO guidance on Copyright and artificial intelligence, UK law distinguishes between AI-assisted works and computer-generated works under CDPA section 9(3), and policy debates continue to test whether that category remains fit for purpose^[5]. Two layers of permission matter in practice. First, training AI models on protected music typically involves copying, engaging CDPA sections 16 and 17 and, where relevant, performers’ rights and sound recording rights, so licences or applicable exceptions must be identified. Secondly, deploying outputs raises further issues, including unauthorised reproduction, impersonation, digital replica harms and misleading presentation to consumers and platforms. In the near term, the most credible answer is not purely statutory but contractual and evidential: datasets, permissions, approvals, labelling and liability must all be documented if music is to remain both innovative and trustworthy, as Herndon’s work illustrates.

NFTs and Provenance
NFTs offer one possible response at the level of provenance and fan identity. Major labels have experimented with music-linked NFT strategies, but in music the most useful role for NFTs is not speculative ownership but authentication: who minted, who held, and what that recorded history says about the relationship between an asset and an authorised artist or project.

That said, authenticity depends on rights, not just tokens. Buying an NFT does not transfer copyright; without a written assignment, the buyer receives only whatever licence the issuer grants^[6]. Under English law, the UK Jurisdiction Taskforce’s 2019 Legal Statement on cryptoassets and smart contracts supports treating cryptoassets as property capable of ownership and transfer, but it also underlines the central distinction that the token is separate from the underlying IP^[7]. For music projects, the crucial instrument is therefore the off-chain licence. It should define permitted uses, prohibit acts such as extracting stems, training a model or minting derivatives, and make clear whether any rights transfer automatically with the token or remain personal to the original buyer. Smart contracts may automate payments once conditions are met, but they do not replace written agreements. Code and contract must work together if rights and royalties are to travel downstream.

Creative Passports: Identity as Infrastructure
If provenance is the receipt, identity is the name on the receipt. Creative Passports link works, collaborators, splits and payment details so that creators can be identified and revenue routed accurately. Through artist Imogen Heap’s Mycelia initiative, the Creative Passport is described as a digital container for verified profile information and identifiers, with the ambition of becoming a usable identity standard for music makers across streaming services, distributors, CMOs and other industry actors^[8].

This kind of identity layer matters because authenticity is not only about whether a piece of music sounds real. It is also about whether the surrounding information is reliable enough to support credit, payment and enforcement. A Creative Passport can hold or link to authoritative information about a creator, their roles on particular works, their identifiers, collaborators, agreed splits and payment details. If platforms and CMOs integrate such systems, the ecosystem is far better placed to route rights and money correctly, reducing stranded royalties and disputes over provenance or credit.

Future Avenues: The Road Ahead
The future of music and AI will depend not just on technological adoption, but on whether ownership, attribution, and permission remain clear across the platforms where music already circulates. Some reports suggest that virtual and augmented reality could deliver a £1.4 trillion boost to the global economy by 2030[9]. Yet, the more immediate point is simpler: as digital environments become more layered, proving who made what, and with whose consent, in AI will only increase. Mixed reality therefore remains relevant, but mainly as part of a wider shift in which music travels across increasingly hybrid formats.

Examples from current practice show how quickly those questions are converging. Holly Herndon used Holly+, for an AI-generated cover of Jolene, creating a motion-capture video featuring Herndon’s 3D avatar as Dolly Parton in a fantasy Appalachia. Additionally, spatial technology such as Ray-Ban Meta Smart Glasses lets users stream video with ambient music, with Meta reducing copyright risk by detecting and muting protected audio. Together, these developments show that if ownership, consent and provenance are not secured around AI now, they will become harder to protect as synthetic media grows more sophisticated and more deeply embedded in music culture.

Conclusion
The question “is this real?” used to be philosophical. In music, it is quickly becoming operational. As AI becomes a tool that pushes the industry toward a new baseline, the most resilient models are likely to be those in which copyright and consent remain recognisable. In those models, rights travel across platforms through robust licensing. Provenance tools provide verifiable records of authorised use, and identity layers supply the metadata that makes all of this possible. Music has always offered refuge, an authored space where emotion, identity and belonging converge. As that refuge expands across streaming, social media, and more immersive digital environments, authenticity must extend beyond sound. It must also reach code, contracts, governance, and infrastructure. Only then can the space between the notes remain a place of belonging, and artists truly own the work they create.

---

[1] PwC, Global Entertainment & Media Outlook 2025–2029. Available at: https://www.pwc.com/gx/en/issues/business-model-reinvention/outlook/insights-and-perspectives.html (Accessed: 02 July 2025)

[2] Lavy, M. Montague. (2001). Emotion and the experience of listening to music: a framework for empirical research. Available at: https://doi.org/10.17863/CAM.61066 (Accessed: 02 July 2025) Pg. 100-102

[3] Copyright, Designs and Patents Act 1988 (UK) (CDPA 1988). Available at: https://www.legislation.gov.uk/ukpga/1988/48/contents (Accessed: 25 July 2025)

[4] The Collective Management of Copyright (EU Directive) Regulations 2016. Available at: https://www.legislation.gov.uk/uksi/2016/221 (Accessed: 25th July 2025)

[5] UKIPO guidance – “Copyright and artificial intelligence”. Available at: https://www.gov.uk/government/consultations/copyright-and-artificial-intelligence (Accessed: 25 July 2025) para 131

[6] Terras, Melissa, Schafer, Burkhard & Favreau, Amélie. (2024). Ownership and control in the creative economy: On new property rights for digital assets. In Terras, Melissa, Jones, Valentine, Osborne, Natalie & Speed, Chris (eds), Data-Driven Innovation in the Creative Industries. Routledge Research in the Creative and Cultural Industries. Routledge. Available at: https://doi.org/10.4324/9781003365891-7 (Accessed: 17 March 2026) Pg. 164

[7] UK Jurisdiction Taskforce (LawTechUK) Legal Statement on Cryptoassets and Smart Contracts (2019). Available at: https://lawtechuk.io/ukjt/legal-statement-on-cryptoassets-and-smart-contracts/ (Accessed: 18 March 2026) Pg. 7, para 15(d); 11, para 33; 26, para 110

[8] Department for Digital, Culture, Media and Sport. (2021). Boundless Creativity: Culture in a time of crisis. GOV.UK. Available at: https://www.gov.uk/government/publications/boundless-creativity-report/boundless-creativity-report (Accessed: 28 May 2025)

[9] PwC, Virtual and augmented reality could deliver a £1.4trillion boost to the global economy by 2030. Available at: https://www.pwc.com/id/en/media-centre/press-release/2020/english/virtual-and-augmented-reality-could-deliver-a-p1-4trillion-boost.html (Accessed: 02 July 2025)