African companies and governments are rushing to adopt AI built in the US and China without the ability to test whether it is safe for their own people; African officials told the UN they want a role in writing the rules.
African companies are racing to put AI to work, and almost nobody is asking whether it is safe. Jonathan Shock at the University of Cape Town puts it plainly: there is essentially no real discussion of AI safety, and so a genuine vacuum.
The damage is already showing up
Kenya used a machine learning algorithm to set health insurance contributions, and it pushed costs up for poor households. Scammers in West Africa used AI-generated content to blackmail victims. Boko Haram bypassed safeguards and used AI to organise attacks and design weapons.
Of 85 large companies PwC surveyed in Africa this year, more than 80% were running AI pilots, but most had no formal structure for overseeing safety. As Mélanie Keïta, who runs a Nairobi financing firm, puts it, everyone is simply trying to catch up.
Who holds the off switch
She has tested both American and Chinese models and says each carries its own risks. Her worry is that Africa becomes a product of foreign corporations that own all its data, rather than being autonomous and free.
Jane Munga of the Carnegie Endowment sharpens the point: ChatGPT and Claude are closed-weight models, centrally controlled by their providers. Chinese models such as DeepSeek and Kimi are open-weight, downloadable and modifiable, which makes them hard for their developers to shut down remotely.
The distinction that matters is not Chinese versus American, she says. It is: who holds the off switch?
A small incident, a catastrophe for Africa
Fewer than half of African countries have an AI policy, yet governments are already using AI in healthcare, education and public services. Kenya is the only African nation in an international AI safety network, and that exercise focused on common global risks, not risks specific to Africa.
Gathoni Ireri, a researcher at the Nairobi-based Ilina Program, argues safety evaluations should judge harm against the size of each country's economy: an incident considered minor in the West could do irrevocable damage in an African country. Companies currently test themselves, she says, and that needs to be replaced by third-party evaluation.
At the UN, Somalia's state minister for foreign affairs, Ali Mohamed Omar, said it outright: 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.
Why it matters
The question of whether to run on Chinese or American AI is the wrong one: what a system is allowed to do in your country depends on who holds the off switch, and who gets to decide what counts as harm. By putting that line on the table, Africa asks it on behalf of every country that imports its technology.
Curated from high-quality sources, with concise summaries and key takeaways.
