Ghostwriters of the Future: Who Owns AI’s Creations in Kenya?

Introduction

From a simple human prompt, Artificial Intelligence (AI) can now author novels, compose music, generate art, and even create stunning audio-video skits, all with just a few words of instruction from a human.  In what was previously imagined to be the exclusive products of human intelligence, AI can now produce artistic work in a fraction of the time. AI’s ability to interpret large data sets, learn from such data and use that processing to achieve specific goals and tasks through flexible adaptation, makes it particularly adept at generating content that rivals human capabilities.

The global impact of AI is undeniable, stretching from developers in Silicon Valley to widespread adoption in countries like Kenya, fundamentally transforming how we create and share content and artistic works. Amid fierce competition to create more advanced AI, the latest generation of AI can now produce a vast variety of content that closely mimics human work. However, as AI-generated content continues to rival human creations, a critical question arises: who holds the copyright to this work?

Where do we draw the line between tool and creator? This article explores this key legal question by addressing what copyright law says about protecting original works, and who gets to claim ownership of AI-made content. This analysis focuses on Kenyan laws in the context of global developments seeking to respond to this challenge.

Which Works are Eligible for Copyright?

Copyright law in most jurisdictions is built on the idea of originality. Originality is not restricted merely to the creation of something new but also to mean that the work must show a human’s skill, effort, and judgement. For instance, if an author writes a poem, their unique choice of words and emotions makes it original, and copyright law protects it as the authors. In Kenya, the Copyright Act (Cap. 130) aligns with this principle by explicitly linking copyright eligibility to originality that stems from human effort.

In particular, the Kenya Copyright Act specifies that for a literary, musical, or artistic work to be eligible for copyright, two conditions must be met:

• sufficient effort must have been spent on making the work to give it an original character; and

• the work must be written down, recorded, or otherwise reduced to a material form.

This “sufficient effort” requirement codifies the principle that copyright protection is a reward for human intellectual labour and judgment. The law isn’t just looking for novelty; it’s looking for the imprint of a human creator’s work.

This human-centric foundation is further reinforced by how the Copyright Act defines the “author” i.e., the creator of the work. The law consistently anchors authorship in a person, even when technology is heavily involved in the creation process. For instance, in the case of a computer-generated literary or artistic work, the author is legally defined as “the person by whom the arrangements necessary for the creation of the work were undertaken”. This is critical because it pre-emptively addresses the question of non-human authorship; instead of granting ownership to the computer or AI, the law assigns it to the human who orchestrated the creation. However, what is deemed “arrangements”, and can a mere prompt qualify?

Global Approaches to the Copyrightability of AI-Generated Content

This is precisely the question our Kenyan law has yet to answer. The nascent and dynamic nature of AI has outpaced legislative developments, and Kenyan courts have not yet had the opportunity to adjudicate on the copyrightability AI-generated works. As a result, there is no judicial precedent or formal legal interpretation to guide us. The definitions of crucial terms like “author”, and qualifications of “sufficient effort” remain theoretical, leaving legal uncertainty that makes it difficult to decisively determine whether a specific AI-generated work qualifies for copyright protection, or if it falls into the public domain.

In the absence of local judicial precent, recent findings from foreign jurisdictions offer valuable insight into how these legal challenges are being addressed. The principle of human authorship was recently tested in the landmark United States case of Thaler vs. Perlmutter,  in which Stephen Thaler attempted to copyright artwork that he had created through AI. In this case, Stephen Thaler had produced a piece of artwork entirely through an AI programme, and the artwork became popular leading to its use by other persons. As a result, Stephen Thaler sought to copyright the artwork as his own, seeking to reap the benefits arising from owning the copyright. He argued that by inputting the specific prompt into the AI which gave rise to the work, he was entitled to copyright it as his own. The U.S. Copyright Office and the court held a different view, finding that the artwork could not be copyrighted, as copyright protection only applies to works made and created by humans. The judge determined that the legal definition of “authorship” refers to human creativity, a standard that work generated solely by AI cannot meet. In reaching this conclusion, the court was guided by other judicial rulings that have consistently held that authors must be human. This affirms a foundational principle: copyright protection is fundamentally tied to human creation. This means that simply giving an AI a prompt, such as “draw a sunset”, is not enough.

However, the court deliberately side-stepped the issue of how much creative input a human must have. Ultimately, it confined its judgement out of judicial restraint, believing its role was to interpret the law as it currently stands, not to create new policy in response to technological advancements. Much like Kenyan law, the U.S. Copyright Act implies a human author by referencing concepts like lifespan, family, and intention to its meaning – qualities that a machine does not possess. It is on this basis that the court denied statutory grounds for extending authorship, hence copyright protection, to a work generated solely by AI.

Determining the Threshold of “Sufficient Effort”

While the decision in Thaler vs. Perlmutter established that a human must be the author, there remains a legal gap that is central to the ever-pertinent question of the copyrightability of AI-generated content. There is still a need to answer how much human input is required for work to be copyrightable.

This unanswered question creates a lot of ambiguity and uncertainty in legal interpretation and statutory enforcement. On one end are works generated with minimal human interaction, which will likely be considered public domain. At the other end are works where a human has substantially modified, selected, or arranged AI outputs, which may qualify for copyright protection. The future of copyright law will be defined by how legislators and regulatory authorities approach this spectrum. They will have to create a legal standard that can distinguish between a human merely acting as an operator and a human exercising genuine creative control, a distinction that will ultimately determine the future of ownership in an age of automated outputs.

Copyright law has an underlying objective to promote the development of science and arts and to improve the livelihood of the creative artists. As such, AI is changing the creative industry by making it easier and faster to create content. However, the lack of in-depth laws that cater to the ever-evolving world of AI also raises tough questions and legal gaps in Kenya and the world over. The country is embracing AI in all industries and sectors, but its copyright laws need to keep pace to protect creators and encourage innovation. The legislature, relevant regulatory authorities, scholars, and experts need to work together to define what “sufficient” human effort looks like.

Recognising this challenge, efforts are underway at both the global and national levels to develop guiding frameworks. At the international level, the World Intellectual Property Organisation (WIPO) is spearheading the conversation, hosting ongoing dialogues and consultations with member states to explore potential policy options and foster a common understanding of how copyright should apply to AI. While no binding international treaty has been established, these discussions are crucial for shaping future global norms. Locally, the Kenya Copyright Board (KECOBO) has been organising stakeholder engagements for the development of a policy framework that addresses the emergence and growth of AI.

Conclusion

While the Kenyan Copyright Act firmly closes the door on AI being a copyright holder, it leaves a critical question open: what constitutes “sufficient effort” to establish human authorship? As AI-generated works become more common, this ambiguity will need to be resolved through Kenyan judicial interpretation, policy guidance, or direct legislative amendment to define where the threshold for original character lies.

The path forward requires a careful balancing act. For Kenya, the goal will be to create a legal framework that incentivises human creativity and protects creators without stifling the innovation that AI facilitates.