There’s something almost surreal about watching artificial intelligence create. You type a few words, press enter, and suddenly there’s an image, a paragraph, sometimes even something that feels deeply thoughtful. It raises a quiet but powerful question: who does this belong to?It sounds like a question for the future, but in Kenya, the answer is already taking shape and it’s surprisingly simple. Strip away all the excitement around AI, and the law keeps coming back to something very familiar: creativity, at least in the eyes of the law, is still human.
The starting point is always the statute. Section 2 of the Copyright Act defines an author as “the person who first makes or creates the work.” It even goes a step further and addresses computer-generated works, saying the author is “the person who arranged for the creation of the work.” That wording feels almost understated, but it does something powerful, it quietly removes the possibility of a machine ever being an author. The law is not interested in the tool; it is interested in the person behind it, so no matter how advanced AI becomes, Kenyan law keeps asking the same question: who, as a human being, created this?
That human focus becomes even more important when you look at what qualifies for protection in the first place. Section 22(3) tells us that a work must have “sufficient effort… to give it an original character.” That phrase, sufficient effort, is where everything begins to slow down. It is not about speed or automation. It is about thought, intention, and creative judgment. When AI produces something on its own, without meaningful human involvement, it becomes difficult to say that anyone truly put in that effort and if no one did, then legally, there is nothing to protect.
This is not just theory anymore. It became very real in the case of Aryeh Movement Limited v. Cynthia Beldina Akoth Okello before the Kenya Copyright Tribunal. The Tribunal was faced with a question that feels very modern: can AI-generated works be copyrighted? Its response was careful, but firm. AI cannot be an author because the law requires authorship to come from a natural or legal person. What stands out in that decision is its restraint. The Tribunal did not try to stretch the law to fit the technology. Instead, it brought the conversation back to something steady and familiar, human input.
There is a point in this conversation where things begin to shift, when someone is no longer just using AI, but actively shaping it. Think of a person who does not just type prompts, but goes further: they retrain a model, fine-tune its behaviour, adjust how it responds, or even build something on top of an existing system. At that stage, the relationship between the human and the machine starts to feel different. It is no longer just interaction; it begins to look like creation and naturally, the question follows: if you have redefined the AI itself, can you claim ownership over what it produces?
Under Kenyan law, the answer starts by separating two things that often get blurred together: the AI system and the content it generates. When someone modifies or develops an AI system, what they are really creating is a form of software and software, under the Copyright Act (Kenya), is protected as a literary work. The law recognises the author of such a work as the person who controlled its development. So, in that sense, yes, if you have genuinely redefined or developed an AI system, you can claim ownership over the system itself. That part of the law is relatively clear and sits comfortably within existing copyright principles.
Though, ownership of the system does not automatically stretch to everything the system produces. This is where the law draws a careful line. Even if you built or trained the AI, the Copyright Act still asks a separate question for every piece of output: who created this particular work? Section 2 insists on a human author, and Section 22(3) requires “sufficient effort… to give it an original character.” That means the law looks at the specific output, not just the tool behind it. If the output is generated without meaningful human input at that stage, it becomes difficult to say that the developer of the AI is also the author of the work.
A useful way to understand this is to step away from technology for a moment. Imagine building a camera. Designing and assembling that camera is a creative and technical achievement in its own right, and the law would protect that work but once the camera is in use, the photographs taken with it belong to the photographer, not the manufacturer. Redefining AI is, in many ways, like building a more advanced camera. You may shape how it works, how it sees, even how it interprets the world but ownership of each image still depends on who actually made the creative choices behind that image.
That said, redefining AI is not irrelevant to ownership, it just does not settle the question on its own. It can strengthen a claim where the person who modified the AI also plays a direct creative role in generating the output. For instance, if someone trains a model in a very specific artistic style and then actively directs, refines, and edits its outputs, their contribution may go beyond mere technical involvement. In such cases, their role begins to look less like that of a toolmaker and more like that of a creator and the law, which is ultimately concerned with human creativity, may recognise that.
If you step outside Kenya for a moment, you begin to realise that this uncertainty around AI and authorship isn’t unique, it’s something courts elsewhere have already had to confront, and quite directly. In Thaler v. Perlmutter, the issue was pushed almost to its breaking point. An image had been created entirely by an AI system, no human shaping it, no editing, no creative intervention at all and then came an unusual move: Dr. Stephen Thaler listed the AI itself, the “Creativity Machine,” as the author, while claiming ownership simply because he owned the system. The law didn’t hesitate, both the U.S. Copyright Office and the courts were clear, almost to the point of simplicity, copyright law protects human creation. Without a human author, there is nothing to protect. It didn’t matter how advanced the AI was, or who controlled it. What mattered was that no person had actually created the work in the way the law understands creation.
Yet, the conversation doesn’t end there. The U.S. Copyright Office adds a more nuanced layer, one that feels closer to how people actually use AI today. It recognises that many works are not purely machine-made or purely human, they sit somewhere in between. So instead of rejecting them entirely, the law begins to separate them. It asks: which parts came from the human, and which did not?. Where AI has generated portions of a work, those parts must be set aside disclaimed but where a person has contributed real creative input, those elements can still be protected. It’s a careful, almost surgical approach to authorship. The law doesn’t deny the presence of AI, but it refuses to let it replace the human creator at the centre of copyright.
When you bring that thinking back into the Kenyan context, it feels surprisingly familiar. Kenyan law is built on the same foundation, that authorship must be human, and that originality flows from intellectual effort. The lesson from Thaler v. Perlmutter isn’t just about one failed claim; it’s about where the law quietly draws its boundaries. Those boundaries aren’t shaped by how advanced technology becomes. They are shaped by something much simpler, and much older: the idea that creation, in the legal sense, is ultimately a human act.
Machines can assist, inspire and even surprise us but they don’t replace the human author and maybe that’s the real point. In a world where it’s becoming easier and faster to create, the law is gently reminding us that creativity isn’t just about producing something. It’s about thinking, choosing, shaping, and intending. It’s about being human.
