The rapid development of artificial intelligence (AI) is reshaping the music industry and redefining how music is created and consumed. Historically, the music sector has had a significant impact on copyright law, meaning that the protection of AI-generated music could have broader implications for copyright policy.
This article explores the implications of AI-generated music, with a focus on copyright law.
AI-generated music employs advanced generative models that use neural networks to analyse large datasets of different genres, styles and compositions, allowing AI to learn musical structures.
The process begins with a large language model that translates data sets into numerical representations, allowing the AI to identify relationships within the data. Integrating music theory into the algorithms ensures that compositions adhere to harmonic and melodic rules, resulting in pleasing and structurally sound music.
pitting neural networks against each other, ensuring more accurate and creative outputs. Expert oversight remains essential to ensure the quality and reliability of the final product.
Platforms such as Suno.com or Udio.com illustrate the application of AI to music production. This accessibility democratises music creation, enabling individuals without formal training to produce music, while providing professionals with tools to streamline production and reduce costs. For users, AI serves as an interactive platform and a valuable educational resource, that fosters creativity.
When assessing the legal risks of employing AI in the music industry, key concerns include intellectual property, liability and ethics. These issues can be summarised in the following table:
| Copyright Issues | Liability Concerns | Contractual Issues | Ethical Concerns | Data Privacy | Moral Rights | AI Regulation |
| Ownership of AI-Generated Music | Accountability for AI actions that infringe third parties ‘rights | Artist and Producer Agreements | Exploitation of AI vs. Protection of Human Creators | Data Use and Privacy Concerns | Moral Rights of Human Artists | Compliance of AI in Creative Fields with the AI Act |
| Use of Pre-existing Works in Training AI | Defamation, Harassment, or Offensive Content | AI Licenses | Transparency and Disclosure | |||
| Copyright Infringement by AI |
For the purposes of this paper, we will focus specifically on the copyright issues.
As an artistic work, music is protected by copyright and related rights. The Berne Convention, reinforced by the WIPO Copyright Treaty and the TRIPS Agreement, requires that works be original and have human authorship to be protected, excluding works with minimal human input.
In the EU, the InfoSoc Directive and the Directive on Copyright in the Digital Single Market (CDSM) provide a harmonised copyright framework for emerging technologies, including AI. The CDSM Directive includes collective licensing and statutory remuneration rights to safeguard human creators in an AI-driven market. Related rights, including those for phonogram producers, provide limited protection for AI-generated output, emphasising investment rather than originality.
At the national level, the UK Copyright, Designs and Patents Act 1988 (CDPA) recognises “computer-generated works” and assigns copyright to the entity responsible for their creation, either the programmer or the user. However, this approach highlights the need for harmonisation across jurisdictions, as most European laws are based on human-centred concepts.
Another relevant aspect is the use of copyrighted material in datasets for AI training without prior permission from the rights-holders, unless an exception applies.
The CDSM provides two exceptions for text and data mining (TDM). Article 3 allows TDM for the purpose of scientific research by research organisations and cultural heritage institutions, provided there is lawful access to the work. However, this does not cover commercial uses typical of AI development. Article 4 allows TDM for other purposes, including commercial uses, unless the rightsholder has expressly reserved their rights (opt-out), as specified in Article 4(3) of the Directive and reinforced by the EU Artificial Intelligence Act. Compliance with these frameworks requires robust licensing agreements or reliance on public domain works to avoid infringement.
Finally, AI systems may inadvertently reproduce copyright-protected music, raising liability concerns. The Court of Justice of the European Union (CJEU) has ruled that even minimal sound sampling requires permission from the rights-holder, unless the sample is unrecognisable or fall within a copyright exception.
Placing fully AI-generated works in the public domain could resolve ownership issues. Machines, unlike humans, do not need the incentives provided by copyright law. This could also reduce the risk of anti-competitive practices arising from excessive monopolisation of AI-generated works. However, denying copyright protection may reduce the incentives to invest in AI technologies.
The introduction of a sui generis right could strike a balance. The scope of such a right should be limited, allowing the owner to prevent others from using and exploiting simple verbatim copies of the AI-generated output. Moreover, such right should be short-lived, lasting three years from publication. The narrower and shorter duration would mitigate the risk of stifling human creativity caused by widespread reliance on AI technologies, while restricting the infringement to verbatim copying would ensure that human creators retain the freedom to use, adapt, and rearrange AI-generated works without legal constraints.
Additionally, clear guidelines are needed for the use of copyrighted works in AI training datasets, requiring permission form rights-holders unless specific exceptions apply, such as those for non-commercial text and data mining.
The emergence of AI-generated music presents both opportunities and challenges for the music industry and copyright law. Existing copyright frameworks, particularly in the EU, may need to be adapted to meet these challenges, ensuring that human creators are protected while encouraging innovation.
Proposed solutions, such as clearer guidelines and the introduction of sui generis rights, aim to balance the protection of human creativity with the unique nature of AI-generated works.
As AI technology evolves, collaboration between legal, technological, and societal stakeholders will be essential to create a fair and sustainable framework for the future of music creation and copyright.
