Rethinking Section 9(3) CDPA 1988 in the Age of Generative AI

Introduction

Recent advances in generative artificial intelligence have placed considerable strain on the conceptual foundations of UK copyright law. Systems capable of producing text, images, and music with minimal human intervention now challenge the long-standing assumption that creative works are necessarily the product of human intellect. Nowhere is this tension more apparent than in section 9(3) of the Copyright, Designs and Patents Act 1988 (CDPA), which provides that, in the case of a computer-generated work, the author shall be “the person by whom the arrangements necessary for the creation of the work are undertaken.”[1] While this provision was originally intended as a pragmatic solution to an emerging technological problem, its continued application in the context of modern generative AI raises serious doctrinal and normative concerns.

Attribution of Authorship

When section 9(3) was introduced, computers were largely understood as sophisticated tools operating under direct human control. The provision was designed to avoid gaps in protection by ensuring that works produced with the assistance of computers did not fall automatically into the public domain.[2] However, this rationale sits uneasily with contemporary AI systems, which do not merely assist human creators but instead generate outputs through complex, autonomous processes. The attribution of authorship under section 9(3) therefore increasingly resembles a legal fiction rather than a meaningful reflection of creative responsibility.

Identifying the individual who undertakes the “arrangements necessary” for the creation of an AI-generated work is particularly problematic. Several actors may plausibly be identified, including the programmer who designs the model, the developer who curates training data, or the user who inputs prompts. Yet none of these individuals exercises direct control over the expressive content of the final output. The programmer does not determine the specific form of any particular work, while the user’s contribution often amounts to a general instruction rather than a creative choice.[3] As a result, section 9(3) attributes authorship without requiring the kind of intellectual engagement traditionally associated with copyright protection.

Inconsistencies in the Law of Originality

This difficulty is compounded by developments in the law of originality. Following the decision of the Court of Justice in Infopaq International A/S v Danske Dagblades Forening, originality requires that a work be the author’s “own intellectual creation,” reflecting free and creative choices.[4] Although the UK is no longer bound by EU law, this standard continues to inform domestic copyright jurisprudence.[5] AI-generated works struggle to satisfy this requirement, as the decisions shaping the final output are made by the system itself rather than by a human author. Section 9(3) thus produces an internal inconsistency: it assigns authorship where the doctrinal basis for originality may be absent.

The implications of this inconsistency are not merely theoretical. By granting copyright protection to AI-generated works through a statutory fiction, the law risks expanding exclusive rights beyond their traditional justification. Copyright has long been defended on the basis that it rewards and incentivises human creativity.[6] Protecting AI-generated outputs, however, may serve primarily to consolidate control in the hands of developers and platform owners, rather than to encourage creative endeavour. In this sense, section 9(3) threatens to distort the balance between private rights and the public domain.

Reform

Reconsideration of the provision is therefore warranted. One possible approach would be to repeal section 9(3), allowing works produced without meaningful human authorship to enter the public domain. This would preserve the human-centred foundations of copyright law while promoting access and innovation. Alternatively, the provision could be retained but applied more narrowly, with protection limited to cases in which a human can demonstrate substantial creative input. A more ambitious reform would involve the creation of a sui generis right for AI-generated works, offering limited protection without equating machine output with human authorship.[7]

Conclusion

Ultimately, section 9(3) reflects an understanding of technology that no longer corresponds to reality. In an era of creative machines, the continued reliance on legal fictions risks undermining both the coherence and the legitimacy of copyright law. Rather than forcing artificial intelligence into existing doctrinal categories, the law must confront the possibility that not all creative outputs require ownership. Recognising the limits of copyright may prove essential to its future relevance.

Refrences

[1] Copyrights, Rights, Patents Act 1988, s 9(3).

[2] Lionel Bently and Brad Sherman, Intellectual Property Law ( 5th edn, OUP 2018.

[3] Andres Guadamaz, ‘Artificial Intelligence and Copyright’ (2017) 28(2) Intellectual Property Ǫuarterly 169.

[4] Inofopaq International A/S v Danske Dagblades Forening (C-5/08) [2009] ECR I-6569 [37]

[5] Football Dataco Ltd v Yahoo! Ltd [2012] EWCA Civ 153.

[6] William Cornish, David Llewelyn and Tanya Aplin, Intellectual Property: Patents, Copyright, Trademarks

[7] UK Intellectual Property Office, Artificial Intelligence and Intellectual Property: Call for Views (2020).

Image Credits: Zach M on Unsplash