Authored By: Prisca E.Hayford
University of South Wales
Abstract
This article explores whether UK copyright law is prepared to address the growing influence of AI-generated creativity. Tools like ChatGPT, Midjourney, and DALL-E are now producing works that rival those created by humans. However, current legislation, especially Section 9(3) of the Copyright, Designs and Patents Act 1988, still focuses primarily on human authorship. By examining relevant laws, case decisions and the UK Intellectual Property Office’s consultations, this article evaluates potential reforms. These range from removing copyright protection for purely AI-generated works to establishing new rights with limited durations. Ultimately, the analysis suggests that UK law needs to evolve to carefully balance encouraging innovation, protecting creators’ interests, and serving the wider public in this era of digital creativity.
Introduction
Recent advances in artificial intelligence (AI) technologies are reshaping the creative industries in ways that would have been unimaginable a decade ago. Generative systems, such as ChatGPT, Midjourney, DALL-E, and Sora, are capable of producing text, images, music, and other forms of content that are often indistinguishable from human creations. For many digital creators, these tools have become a core part of the creative process, enabling new forms of expression and production at an unprecedented speed and scale.
However, this technological leap has also brought significant legal challenges. Chief among them is the question of authorship: when a work is generated wholly or partly by AI, who – if anyone – can claim the rights to it? UK copyright law, which was drafted with human creators in mind, now faces pressure to adapt to a world in which non-human agents contribute to, or even independently generate, creative works. The debate extends beyond mere legal technicalities, touching on broader issues of innovation, economic incentive, and the protection of human artistry in an increasingly automated world.
UK Copyright Law Overview
In the United Kingdom, copyright protection is governed by the Copyright, Designs and Patents Act 1988 (CDPA). To qualify for protection, a work must meet three fundamental criteria. First, it must be original, meaning it reflects the author’s intellectual creation and involves the application of skill, labour, or judgment.[1]. Second, it must be fixed in some form that allows it to be reproduced[2]. Third, it must have an identifiable author, generally the person who created it[3].
A distinctive element of the UK framework is section 9 (3), which deals with the “computer-generated works”. In these cases, the law treats the author as the person by whom the arrangements necessary for the creation of the work are undertaken[4]This provision was drafted at a time when digital creativity was limited to relatively basic, rule-based programs. It was never designed to address the capabilities of modern generative AI systems, which can produce highly complex and seemingly original outputs with minimal human intervention.
The UK Intellectual Property Office (IPO) has acknowledged that section 9(3) may be increasingly out of step with technological realities.[5]. Its recent consultations have considered several reform options, including removing protection for purely AI-generated works altogether or creating a separate right with a shorter duration of protection.[6].
Other jurisdictions take a different stance. In the United States, the Copyright Office maintains that works lacking human authorship cannot be copyrighted.[7], a position upheld in Thaler v Perlmutter[8]An AI-generated image produced by the “Creativity Machine” was refused registration. The European Union similarly ties authorship to human creators, but has introduced exceptions for text and data mining (TDM) under the Digital Single Market Directive to support AI development.[9]. These allow certain automated uses of copyrighted works for research and in some cases, commercial purposes, subject to safeguards for rights holders.
AI’s Creative Abilities and Legal Challenges
Artificial intelligence is no longer confined to analytical or repetitive tasks; it is increasingly capable of producing content that resembles (and sometimes rivals) the work of human creators. Modern systems can compose music, generate photorealistic images, write poetry, and even summarise complex legal judgments. In some instances, these outputs are produced with minimal human input beyond providing a short prompt.
From a legal standpoint, this raises fundamental questions about the application of existing copyright principles. Traditionally, authorship is linked to human creativity, intention, and intellectual effort.[10]. An AI system, however, does not possess consciousness or intention in the legal sense, making it difficult to fit within these established concepts[11].
This creates a practical and philosophical divide: should AI be regarded merely as a sophisticated tool under the control of a human user or as an autonomous creator in its own right? If the former, the human operator might claim copyright based on having exercised sufficient skill, labour, or judgment in producing the work[12]. If the latter, current UK law offers limited solutions, as section 9(3) of the CDPA attributes authorship to “the person by whom the arrangements necessary for the creation of the work are undertaken[13]”, without defining the threshold of human involvement required.
Internationally, similar tensions are visible. In Thaler v Perlmutter, the US Copyright Office rejected registration for an AI-generated artwork, reaffirming that copyright subsists only in works with a human author.[14]. The European Union has maintained its human-authorship requirement but has expanded text and data mining (TDM) exceptions to facilitate AI development, reflecting a policy choice to encourage innovation while preserving core authorship principles[15].
The legal uncertainty has real-world implications. Without clear rules, disputes over ownership of AI-generated works could deter investment, while overly broad protection risks undermining the rights of human artists whose work may be used, often without consent, to train AI systems.[16]. Balancing innovation with fair protection remains one of the most pressing challenges in contemporary copyright law.
Case Law and Policy Developments
In recent years, legal and policy discourse in the UK has increasingly focused on the implications of AI-generated works for copyright law. A central reference point is the UK’s Intellectual Property Office’s (IPO) series of consultations and reports between 2021 and 2022, which sought stakeholder views on whether the Copyright, Designs and Patents Act 1988 (CDPA) should be reformed to address AI authorship. The IPO examined whether the current “computer-generated works” provision in section 9(3) continues to provide an appropriate balance between incentivising AI-driven creativity and protecting the interests of human authors. Responses revealed a sharp divide: some stakeholders argued that retaining copyright for computer-generated works fosters investment in AI innovation, while others contended that it risks granting monopolies over content created without human intellectual effort.[17].
A particularly influential case in this debate is Thaler v Comptroller-General of Patents, Designs and Trade Marks, where the Court of Appeal reaffirmed that UK patent law requires a named human inventor.[18]. Although Thaler contained patents rather than copyright, its reasoning has wider implications: it demonstrates judicial reluctance to recognise non-human entities as authors or inventors under existing intellectual property frameworks. In Thaler, the applicant sought to have his AI system “DABUS” named as inventor on two patent applications. The court held that only a “natural person” could be an inventor within the meaning of the Patents Act 1977, thereby rejecting the notion of AI legal personhood in this context.
Internationally, similar positions have been adopted. In the United States, the Copyright Office has refused registration for works created solely by AI without human authorship, as in Thaler v Perlmutter, where the Office rejected a request to register an AI-generated artwork. Its position, confirmed in its 2023 policy statement, is that copyright subsists where there is “human authorship” in the traditional sense.[19]. The European Union, while yet to legislate specifically on AI authorship, has maintained through its acquis that originality requires the authors ‘own intellectual creation, implicitly excluding works without human creative input.[20].
The IPO’s 2022 government response highlighted three possible routes for reform: (1) removing protection for purely computer-generated works, (2) replacing the current provision with a right of reduced scope or duration, or (3) maintaining the status quo. This policy work reflects a recognition that technological capability is outpacing statutory design and that future UK copyright law will need to reconcile protection with accessibility in a digital economy where AI is an active creative participant.[21].
Core Legal Tensions
At the heart of the AI copyright debate lies the fundamental question: can works generated without human mental effort meet the legal threshold of “originality”? Under UK law, originality generally requires that the work is the author’s intellectual creation, involving skill, labour, and judgment.[22]. When AI autonomously produces outputs, there is no human mind exercising creative choice in the conventional sense. This creates a tension between the de jure requirements of copyright and the de facto capabilities of modern AI.
Section 9(3) of the CDPA attempts to address this by attributing authorship of computer-generated works to the “person by whom the arrangements necessary for the creation of the work are undertaken”[23]. However, this position, drafted in 1988, was conceived in an era when “computer-generated” meant works produced with substantial human instruction. With generative AI models such as GPT-4, Mid-Journey, and Sora capable of producing complex and original content with minimal human input, critics argue that section 9(3) is ill-equipped for the realities of 21st-century creativity.[24].
There are also ethical and economic dimensions to this tension. One concern is the unconsented use of human-created works in AI training data sets, which may amount to mass, unlicensed copying. This has led to litigation in multiple jurisdictions, with claimants arguing that such practices undermine the livelihood of human creators while enriching AI developers.[25]. Furthermore, there is apprehension that granting full copyright to AI-generated works could crowd out human artistry in cultural and commercial markets, especially if AI could generate high-quality content at a negligible marginal cost.[26].
The debate intersects with broader discussions about whether AI should be granted a form of legal personhood. In a recent discussion paper, the Law Commission explored whether highly autonomous AI systems might, in the future, be recognised as having a distinct legal status for certain purposes – a notion that could radically alter the basis of intellectual property law.[27]. While the Commission stopped short of recommending AI personhood, its acknowledgment of the concept reflects an awareness that technological advances could challenge core legal categories, including authorship.
Reform Proposals
Reforming UK copyright law to address AI-generated works requires striking a delicate balance between incentivising innovation and safeguarding the interests of human creators. Several potential approaches have emerged in recent policy discussions.
- Abolish protection for purely computer-generated works.
One option, proposed in the UK IPO’s 2021-2022 consultations, is to remove section 9(3) entirely, restricting copyright to works with a human author.[28]. This would ensure that copyright law remains grounded in human creativity while allowing AI-assisted works to remain protected if sufficient human intellectual input is demonstrated[29]. Critics of this approach argue it could disincentivise investment in AI-generated content, particularly in sectors such as media, design, and advertising, where machine-created works are becoming commercially significant[30].
- Introduce a new, limited right for AI-generated works
A middle-ground proposal is to replace section 9(3) with a sui generis right of reduced duration (for example, five years) for works generated without a human author[31].
The aim would be to acknowledge the economic value of AI outputs while avoiding long monopolies that might stifle cultural reuse. Such a right could be coupled with transparency obligations, requiring disclosure when a work is predominantly AI-generated[32].
- Recognise joint-authorship for human AI collaborations.
Given that many creative processes involve a combination of human input and AI assistance, another reform pathway is to expand joint authorship provisions to cover situations where an AI makes a material creative contribution.[33]. This would avoid the “all-or-nothing” problem inherent in current law, where AI involvement may exclude protection unless a human’s intellectual creativity meets the originality threshold.
- Introduce transparency and ethical safeguards.
Alongside substantive rights, reforms could require AI developers and users to disclose training datasets and ensure compliance with data protection and copyright clearance rules.[34]. Such measures could address ethical concerns about unlicensed copying of human works and help maintain trust between creators and AI innovators.
- International harmonisation
Any UK reform will need to account for developments in other jurisdictions. The EU has introduced text and data mining (TDM) exceptions with an opt-out for rights holders.[35], while the US Copyright Office has reaffirmed that works without human authorship are ineligible for copyright protection[36]. Divergence could create legal uncertainty for cross-border collaborations and trade. As such, the UK may wish to coordinate reforms with its major trading partners to avoid creating incentives for forum shopping.
Ultimately, the optimal approach may be a hybrid system: maintaining human-centered copyright for traditional works, introducing a tailored right for purely AI-generated outputs, and embedding ethical safeguards into both regimes. Such a system would preserve the foundational values of copyright while adapting to the realities of a digital creator revolution.
Conclusion
The UK stands at a pivotal moment in determining how copyright law will adapt to the realities of AI-generated creativity. The rapid evolution of generative AI has exposed gaps in the Copyright, Designs and Patent Act 1988, particularly section 9(3), which was drafted from an earlier era of “computer-generated works” and now faces conceptual and practical strain. While current law remains grounded in human authorship, the economic and cultural significance of generated outputs demands a nuanced and future-proof approach.
The policy choices available range from abolishing protection for purely AI-generated works to creating entirely new categories of rights. Each carries trade-offs: too little protection risks discouraging innovation and disadvantaging human creators. Crucially, any reform must balance the UK’s role as a hub for creative industries with the need to align internationally, especially with the EU and the US, to avoid legal fragmentation.
Ethical considerations should also be at the heart of reform. Transparency in AI training data, robust attribution practices, and safeguards against exploitation of human creators will be essential to maintain trust in creative markets. Without these, legislative changes risk prioritising commercial incentives over artistic integrity.
Ultimately, the “digital creator revolution” is not a passing trend. It is the next phase of cultural reproduction. Whether AI becomes a recognised co-creator, or whether the UK develops a distinct system of rights for machine-generated works, the need for reform is both urgent and inevitable. The challenge for lawmakers will be to design a copyright framework that is adaptable, equitable, and capable of fostering creativity, human and artificial alike, for decades to come.
Reference(S):
[1] University of London Press Ltd v University Tutorial Press Ltd [1916] 2 Ch 601 (Ch). [2]Copyright, Designs and Patent Act 1988, s 3(2).
[3]ibid s 9(1).
[4]ibid s 9(3).
[5] UK Intellectual Property Office, Artificial Intelligence and Intellectual Property: Copyright and Related
Rights (2021) para 32
https://www.gov.uk/government/consultations/artificial-intelligence-and-ip-copyright-and-pat ents/outcome/artificial-intelligence-and-intellectual-property-copyright-and-patents-governm ent-response-to-consultation accessed 7th August 2025.
Prisca Hayford
[6]ibid paras 46-54.
[7] US Copyright Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence.
[8]Thaler v Perlmutter (US Copyright Office, Review Board Decision, 14 February 2022). [9] Directive (EU) 2019/790, arts 3-4.
[10] University of London Press Ltd (n1).
[11]Jane Ginsburg, ‘People Not Machines: Authorship and What It Means in the Berne Convention’ (2018) 49 IIC 131.
[12]Ladbroke (Football) Ltd v William Hill (Football) Ltd [1964] 1 WLR 273 HL.
[13]Copyright, Designs and Patent Act 1988, s 9(3).
[14]Thaler v Perlmutter (DDC, 18th August 2023) Civ Action No 22-1564.
[15] Directive (EU) (n 9).
[16] UK Intellectual Property Office, Artificial Intelligence and Intellectual Property: Copyright and Related Rights (2021) para 68.
[17] UK Intellectual Property Office (n 5).
[18]Thaler v Comptroller-General of Patents, Designs and Trade Marks [2021] EWCA Civ 1374.
[19] US Copyright Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence (16th March 2023) https://www.copyright.gov/ai accessed 9th August 2025.
Prisca Hayford
[20] Directive 2001/29/EC of the European Parliament and of the Council of 22 May 2001 on the harmonisation
of certain aspects of copyright and related rights in the information society [2001] OJ L167/10, art 2; Case C-145/10 Painer v Standard VerlagsGMBH [2011] ECR I-12533.
[21] UK Intellectual Property Office, Government Response to Consultation on Artificial Intelligence and Intellectual Property (2022).
[22] University of London Press Ltd (n 1).
[23]Copyright (n 13).
[24] UK Intellectual Property Office (n 5)
[25]Getty Images (US) Inc v Stability AI Ltd (SDNY, filed 3 February 2023); Andersen v Stability AI Ltd (ND Cal, filed 13 January 2023).
[26] Andres Guadamuz, ‘Artificial Intelligence and Copyright’ (2020) 39(2) EIPR 59.
[27]Law Commission, Discussion Paper: Digital Assets and Emerging Technologies (2024) para 5.43 https://lawcom.gov.uk/project/digital-assets/ accessed 9th August 2025.
[28] UK Intellectual Property Office (n 21) para 64.
[29]Copyright (n 13).
[30] Andres Guadamuz, ‘Artificial Intelligence and Copyright’ (2020) 39(2) EIPR 59,64.
[31] UK Intellectual Property Office, Artificial Intelligence and Intellectual Property: Copyright and Related Rights (2021) para 57.
Prisca Hayford
[32]Ibid para 59.
[33]Copyright, Designs and Patents Act, s 10.
[34] House of Lords Communications and Digital Committee, Large Language Models and Generative AI (HL Paper 69, 2023-24 ) para 104.
[35] Directive EU 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market [2019] OJ L130/92, arts 3-4.
[36] US Copyright Office (n 19).
Bibliography
Legislation and Policy Sources
UK Intellectual Property Office, Artificial Intelligence and Intellectual Property: Copyright and Related Rights (2021)
https://www.gov.uk/government/publications/artificial-intelligence-and-intellectual-property accessed 9 August 2025
UK Intellectual Property Office, Government Response to Consultation on Artificial Intelligence and Intellectual Property (2022)
https://www.gov.uk/government/consultations/artificial-intelligence-and-intellectual-property -call-for-views accessed 9 August 2025
International Sources
US Copyright Office, Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence (16 March 2023) https://copyright.gov/ai accessed 9 August 2025
World Intellectual Property Organization, WIPO Conversation on Intellectual Property and
Prisca Hayford
Frontier Technologies (2023) https://www.wipo.int/about-ip/en/frontier_technologies accessed 9 August 2025
Academic and Secondary Sources
Law Commission, Artificial Intelligence and the Law: Discussion Paper (2024) European Commission, Copyright in the Digital Single Market (Directive (EU) 2019/790)
