A few weeks ago, African leaders stood at the United Nations Security Council and said something that I think needs to be heard and discussed far beyond the chamber where it was said.
“Africa cannot remain merely a market for systems developed elsewhere, a source of raw data, or a testing ground for a technology it did not help govern,” said Ali Mohamed Omar, Somalia’s state minister for foreign affairs. “Africa must be granted genuine representation and a voice in these multilateral processes.”
Lewis Garseedah Brown II, permanent representative of Liberia, puts it in terms I find even more straight to the point: Africa must be an “equal co-architect in determining the standards, ethics, and architectures of this technological era.”
A Rest of World investigation published last week by Danai Nesta Kupemba documents what is driving these demands. Incidents of AI misuse and malfunction are growing across Africa. Kenya’s predictive machine learning algorithm for setting health insurance contributions has driven up costs for poor households. Scammers in West Africa are using AI-generated content to blackmail victims. Terrorists have used AI to organise attacks and design weapons after bypassing safeguards. More than 80% of large companies surveyed by PricewaterhouseCoopers this year are running AI pilots, many without formal structures to oversee safety. Fewer than half of African countries have AI policies or strategies. Kenya is the only African nation participating in an international AI safety network.
Here is what I think about this article, and it is a point Jonathan Shock, associate professor of mathematics at the University of Cape Town, makes with clarity: “There’s essentially no real discussion about AI safety. So, there’s a real vacuum.”
But I don’t believe that vacuum is accidental!!! AI safety evaluations are currently conducted by the companies building the systems, and they are designed for global deployment, not for the specific contexts, economies, languages, cultures, and harm thresholds of African nations. Gathoni Ireri, a research scholar at the Nairobi-based Ilina Program, said an incident that might be considered minor in a Western country could irrevocably harm an African one. Africa-centric safety evaluations would judge harm by the size of each country’s economy and the specific risks to its population.
Jane Munga of the Carnegie Endowment for International Peace makes another point. The question that matters about AI models, she asks, is: who holds the off switch?
That is a governance question. And the answer, right now, is not African governments or communities or researchers.
Genuine co-authorship of AI safety standards requires the institutional capacity, technical infrastructure, and political standing to use that seat when it matters. Africa must be a co-architect in AI safety standards. What would that really require, and where should it start? Would love to hear your thoughts in the comments!