Kenya's Copyright Tribunal has issued what appears to be the continent's first ruling squarely addressing artificial intelligence and authorship, holding in Aryeh Movement Limited v. Cynthia Beldina Akoth Okello (COPTA/E001/2025) that works generated solely by AI do not qualify for copyright protection under the Copyright Act. The decision leaves open whether AI-assisted works involving genuine human input can still qualify for protection. This is a question which African legislatures are yet to settle.

The ruling crystallises a problem that has been building across the continent: every major African intellectual property statute assumes a human creator or inventor, and none was drafted with generative AI in mind. Copyright regimes rooted in the droit d'auteur tradition, including OAPI's Bangui Agreement, vest rights only in "the natural person who created the work." South Africa's Copyright Act defines authorship of computer programs by reference to human control but, unlike the UK, has no provision for wholly computer-generated works. Nigeria's 2022 Copyright Act defines an author simply as "the creator of a work" without specifying humanity; an ambiguity that researchers say could plausibly be read either way by a Nigerian court.

Patent law faces a parallel gap. ARIPO's Harare Protocol defines an invention as "an idea of an inventor," language that presumes a human agent without expressly requiring one. Across African jurisdictions, the doctrine of conception (which is a definite idea taking shape in a human mind) cannot straightforwardly accommodate an AI system that originates the inventive concept itself, and employer rights typically attach only once a qualifying human inventor has already been identified.

This article argues that African states should not simply import solutions from the United States, the European Union or the United Kingdom. US courts, in Thaler v. Vidal and Thaler v. Perlmutter, have rejected AI inventorship and authorship outright, while the UK's four-decade-old "computer-generated works" provision supplies an answer without asking whether it is the right one. All three jurisdictions, the analysis notes, are also home to AI developers with a direct commercial interest in the outcome [an interest most African states do not share], since they remain consumers rather than producers of AI infrastructure.

The stakes are sharpened by a separate concern: leading generative AI systems have trained extensively on African music, literature, oral tradition and visual art, much of it scraped without consent, raising questions about compensation that mirror longstanding disputes over biopiracy of genetic resources.

The article therefore proposes three concrete steps. ARIPO could issue a harmonisation guideline setting tiered protection based on the depth of human creative or inventive contribution, without requiring legislative amendment by member states. Nigeria and Kenya, described as the continent's most active digital creative markets, could adopt a modified UK-style provision conditioned on disclosure that training data did not involve unconsented use of African works. And the African Union's Digital Transformation Strategy, alongside the AfCFTA Intellectual Property Protocol, could anchor a continental benefit-sharing framework modelled loosely on the Nagoya Protocol's consent regime for genetic resources.

This article cautions that any specific recommendation risks obsolescence within a few years given the pace of the technology. We contend, in conclusion, that the underlying stakes are not time-limited: absent a deliberate departure from frameworks designed for Northern economies, African intellectual property law risks once again defaulting to rules that serve foreign developers rather than the continent's own creators and inventors.