AI copyright policy considerations for Botswana and South Africa – Compensation for starving artists feeding generative AI

The balancing act which domestic intellectual property policy is now challenged to strike is between fostering growth in technological innovation and incentivising creative labour. Ordinarily, these two considerations should not be mutually exclusive, but generative artificial intelligence (Gen AI) has led to a line being drawn in the sand. On one hand, AI undeniably has the potential to exert a profound societal impact, an outcome which shouldn’t be unnecessarily stifled, as onerous barriers to data access would undermine the ideal objectives of copyright policy. On the other hand, creative labourers are still subject to the inherent peculiarities of their industry, which faces further disruption despite their works forming part of Gen AI training data. A few cases worldwide illustrate this standoff, wherein artists contend that automation threatens to disrupt an already precarious creative industry and seek, inter alia, compensation for the use of their works. In light of these polarised viewpoints, a socio-legal and tech-neutral approach is required to analyse whether a balance could be struck in light of Botswana and South Africa’s copyright policies and whether artists could be compensated for this type of use. The legal basis for a claim of compensation will be ascertained first and shall entail determining the manner in which Gen AI infringes upon artists’ exclusive rights, and utilise protected works in a manner which would ordinarily warrant remuneration. This analysis shall be informed by a discussion on the polarising interests involved, the normative justifications for said copyright protection, and the corresponding rationales for its limitations. Before concluding, this article shall evaluate the feasibility of recommendations for fair remuneration frameworks in the context of Botswana and South Africa. It will be suggested that, ideally, CMOs ought to utilise metadata and watermarking measures to trace their members’ works across training data sets and ultimately claim compensation on their behalf. However, it is ultimately found that said CMOs in these countries ought to drastically increase their technical ability to monitor online acts of infringement.