AI-Generated Content and Copyright in Kenya: Who Owns What KECOBO Protects

Kenya’s Copyright Act was not written with generative AI in mind, but it turns out to already contain a provision that speaks directly to it, and a recent Copyright Tribunal ruling has now tested what that provision actually means in practice. For a startup building products around AI-generated content, whether that is marketing copy, illustrations, music, or code, the question of who owns the output, and whether anyone owns it at all, has a real answer under existing Kenyan law.

The provision already sitting in the Copyright Act

Section 2 of the Copyright Act (Cap. 130) defines “author” separately for different categories of work, and one of those categories is directly on point: for “a literary, dramatic, musical or artistic work or computer program which is computer generated,” the Act defines the author as “the person by whom the arrangements necessary for the creation of the work were undertaken.” This is textually almost identical to section 9(3) of the UK’s Copyright, Designs and Patents Act 1988, and it means Kenyan law does not treat a computer-generated work as authorless by definition; it routes authorship to whichever human or corporate person made the arrangements for the work’s creation, not to the machine.

Why “sufficient effort” still has to be human effort

That authorship provision does not operate in isolation. Section 22(3) makes a work eligible for copyright only where “sufficient effort has been expended on making the work to give it an original character,” and where the work has been reduced to material form. A Copyright Tribunal ruling, in a dispute between Aryeh Movement Limited and Cynthia Beldina Akoth over registration of literary works allegedly involving an AI illustrator, addressed this directly: the Tribunal held that section 22(3)’s originality requirement presupposes human creative input, and that works generated autonomously by AI, without demonstrable human creative effort, do not qualify for copyright. Simply typing a prompt and using whatever the AI produces, in the Tribunal’s own framing, is not enough. We were not able to independently confirm the exact date of this ruling from a primary source during this research, since secondary reporting gives inconsistent dates; treat the case as a real and directly relevant Tribunal decision, but confirm its precise date before citing it as authority in any filing.

KECOBO’s jurisdiction has limits too

The same ruling addressed a second, more procedural point: the Tribunal held that KECOBO itself does not have jurisdiction to unilaterally determine authorship or ownership disputes through its registration process. That determination belongs to courts and the Tribunal, not to the Board acting alone. For a business that has registered, or is considering registering, a work with AI involvement, this matters practically: a KECOBO registration is not the forum where a genuine dispute over how much of the work was human-authored gets resolved, and a registration should not be treated as settling that question if it is later challenged.

The practical line: tool versus author

Putting the statute and the Tribunal ruling together, Kenyan law currently draws a workable, if untested-at-appellate-level, line. Fully autonomous AI output, with no identifiable human creative contribution, has no author under any limb of section 2, fails the section 22(3) originality test, and is not protected. Human-directed, AI-assisted work has two independent routes to protection: the ordinary definition of an author as the person who first creates the work, where a human supplies the creative expression and uses AI as a drafting tool; or the computer-generated-work definition, where a human sets the arrangements, prompts, parameters, and selection that shape what the AI produces. Either route still has to clear the section 22(3) bar, which the Tribunal has now read as requiring genuine human creative effort. The practical implication for a business: document the human creative decisions involved in producing AI-assisted content, the editing, curation, selection, and rejection of AI drafts, since that record is what would support a claim to authorship if the point is ever challenged.

A gap the pending Copyright Bill may make worse, not better

A Copyright and Related Rights Bill, 2026, led by KECOBO, was circulated for public comment with a deadline of 31 March 2026 and would substantially rewrite the Act. On the evidence available, the draft narrows the definition of “author” to “a natural person” only, which would remove the existing computer-generated-work limb discussed above rather than clarify it. A civil-society submission on the draft has specifically flagged this as a significant omission, since it would leave human-AI collaborative work in a less certain position than it occupies under the current Act, not a more certain one, and has proposed adding express definitions for AI-generated and AI-assisted work instead. Separately, a Kenya Intellectual Property Bill, 2026 would restructure Kenya’s IP institutions entirely, merging KECOBO into a single new authority, but does not itself address AI-generated content. Neither bill has been enacted; both remain in the legislative process, and businesses should not assume either will resolve today’s uncertainty in their favour.