Generative Artificial Intelligence and Intellectual Property: Training, Authorship and the Fair Allocation of Value in African Creative Industries Under the Nigeria Copyright Act 2022

Abstract

Generative artificial intelligence (AI) is trained on the work of human creators and increasingly competes with it. This raises three connected copyright questions: whether training on copyrighted works infringes, whether AI outputs can be protected and how the value generated by AI should be shared. For African creative industries, notably Nigerian film and music, the questions are acute, because creators have long struggled to capture the value of their work while states also seek to build domestic AI capability. Existing scholarship tends to address training and output separately, leaves the allocation of value implicit and has given limited attention to Nigeria. This study addresses those gaps. Its objectives were to examine the three questions comparatively, to evaluate the position of the Nigeria Copyright Act 2022 and to propose reforms suited to Nigeria and other African states. The study adopted a doctrinal and normative research design with a comparative element. It analysed primary legal sources, including the Nigeria Copyright Act 2022, the Berne Convention and TRIPS Agreement, European Union, United Kingdom and Japanese legislation and leading United States decisions, together with a purposive sample of 28 peer-reviewed journal articles published between 2023 and 2026, each verified against its Crossref publisher record and screened for retractions. The analysis finds that the three questions are interdependent and that answering any one in isolation produces outcomes that are unfair to creators or obstructive to innovation. Nigerian law is at once too restrictive for domestic developers, because its exceptions do not clearly cover training, and too weak against foreign extraction of Nigerian works. It also contains no mechanism for sharing AI's value with creators. The study recommends a conditional text and data mining exception tied to lawful access, a machine-readable opt-out and transparency about training data; a statutory right to equitable remuneration administered through reformed collective management organisations; retention of the human authorship requirement; supplementary protection for voice, likeness and traditional cultural expressions; and regional coordination under the African Continental Free Trade Area.