New Working Paper ‘Copyright and African NLP: What we can learn from the practices and activities of Africa-focused natural language processing’
Data Science Law Lab is pleased to announce the first entry in our working papers released in 2026: ‘Copyright and African NLP: What we can learn from the practices and activities of Africa-focused natural language processing’, by Chijioke Okorie. This paper was presented at the 44th Annual Congress of the of the International Association for the Advancement of Teaching and Research in Intellectual Property, (ATRIP), June 28-July 1, 2026, hosted by the University of São Paulo, Brazil.
________________________________________________________________________________________
Copyright and African NLP: What we can learn from the practices and activities of Africa-focused natural language processing.
Chijioke I Okorie
DSLL Working Paper Series 2026/01
AI innovation across the African continent is on the rise and the number of projects, developers and start-ups creating African language and context-specific datasets, models and and other tools, is increasing. Although this development is clearly welcome and much sought after, it raises concerns shared in other parts of the world regarding the legality of training AI on existing copyright-protected materials and on personal data obtained wrongfully or without consent, as well as the use of biased data that produces misinformation and decisions that infringe human rights. Several African countries have joined their other regions to consider how bespoke legislation or reform of existing legal frameworks including copyright law, might control these risks and harms and rebalance power and positions.
Far less attention, however, has been paid to the actual practices of developers building AI solutions for African contexts: the tools they develop, the data sources they draw on, the licences they choose, the funders and supporters who shape their work, and the repositories and platforms on which they share their work. These are interesting particularly for revealing evidence-led regulatory pathways to emerging issues. This paper reports pilot empirical work examining a sample of African NLP research papers published between 2019 and 2024 along these dimensions: the type of resource created, the data sources used to create them, the licences attached, the funders and supporters behind the work, and the repositories on which they appear. It presents preliminary findings, notably that the data sources behind African NLP are predominantly copyright-protected materials from news media and personal information; that the resources produced are themselves frequently released without clear or explicit licences; and that an overwhelming number of hosting and funding and investments for this work is from outside the continent.
Full paper is available for download.
