In July 2021 South Africa became the first jurisdiction in the world to grant a patent where the inventor is not human, but an artificial intelligence (AI). My doctoral student, Meshandren Naidoo, and I wrote an article on this historic world-first for South Africa.

The AI creativity machine in question, called DABUS, which was made by American entrepreneur Dr Stephen Thaler, invented a new food container. Thaler applied for a patent on this invention in various jurisdictions around the world. During 2020, Thaler’s patent applications were refused by the by European Patent Office (EPO) and the United States Patent and Trademark Office (USPTO). These patent offices followed a strictly textual approach in interpreting their respective patent legislation – thereby excluding AI from the definition of ‘inventor’. In this light, the decision by the South African Companies and Intellectual Property Commission (CIPC) to grant a patent for the same application that was refused by two of the world’s leading patent offices came as a surprise to many. In our article, we investigate whether the South African decision was legally correct. Our analysis is informed by the ground-breaking judgment by the Australian Federal Court on DABUS’s food container patent application. This judgment was handed down just a few days after the South African decision was published and provides sound legal rationales in support of allowing AI inventorship. Importantly, the Australian Federal Court adopted a purposive approach to interpreting patent legislation – in contrast with the EPO and USPTO. Since South African law adheres to a purpose approach, the Australian judgment offers a more suitable precedent to follow. It is concluded that the South African CIPC decision was legally correct. Furthermore, it is clarified that allowing AI inventorship does not necessarily imply that AI has legal personhood. 

Read the full article here.

Like a fox in a chicken coop, the South African Protection of Personal Information Act (POPIA) has caused consternation among some scientists, bioethicists, and biolawyers. Until recently, the issue of acceptable modes of informed consent seemed to have been settled on a position that either specific, broad, or tiered consent is acceptable in South Africa. By contrast, POPIA requires consent to be specific, and for such specific consent to be obtained for various actions that are part of the health research, such as storage and sharing of research information. This has raised concerns that health research in South Africa and with South African collaborators may be impeded. One concern, in particular, is about large biobanks of specimens and data that were build up in the past without specific consent – can these specimens and data still be used for new research projects, or would research participants who donated years ago need to be traced and ‘reconsented’?

Dr Beverley Townsend and I show that POPIA itself offers possible solutions to these problems. Moreover, we show that the possible solutions offered by POPIA can, for the most part, be embedded in South Africa’s well-established and comprehensive regulatory framework for health research. The perceived fox may actually be a guard dog. 

Read our article here.

You can also see our other research on POPIA here.