Think of a 22-year-old first-time voter in Kenya, scrolling through her phone a few months before the 2027 election. The posts look like ordinary people talking. Some of them are not people at all. Behind the machine-made voices there may be no crowd, only a single operator and a cheap tool.
That scenario is the quieter side of Africa’s AI story. The louder side is infrastructure. Africa holds roughly 18% of the world’s population, yet data from the Africa Data Centres Association and the World Economic Forum put its share of global data centre capacity at about 0.6%, against nearly 45% for the United States. A 2026 industry study adds that outside South Africa, only about a third of built capacity is fully used.
But the gap that gets the headlines is not the only one. The deeper gap is over who gets a real say in what these technologies are for, and where their limits lie.
   A rush to write the rules
When the African Union adopted its Continental AI Strategy in July 2024, only six countries (Algeria, Benin, Egypt, Mauritius, Rwanda and Senegal) had national AI strategies of their own. Across the world, 67 countries had already drafted national frameworks between 2017 and 2023.
Now the AU’s first implementation phase, running from 2025 to 2026, has governments moving quickly. Their drafts lean heavily toward economic growth, investment and competitiveness, ahead of a midterm review scheduled for 2027.
    Whose voices are missing
The same urgency meets an uncomfortable reality about language. A 2025 study of major language models found support for only 42 of Africa’s roughly 2,000 languages. Just four (Amharic, Swahili, Afrikaans and Malagasy) get consistent coverage. That leaves more than 98% of the continent’s languages unsupported, which researchers call “language data flaring.”
Kenya shows what is at stake. About 40% of registered voters are aged 18 to 34. Trust Lab researchers identified 17 major coordinated disinformation campaigns in 2025 alone. Anthropic’s recent misuse report described the removal of an account that one actor used to mass-produce Kenyan political content. It is a reminder of how cheaply AI can manufacture the appearance of public consensus before the 2027 general election.
   A small council with a plain question
Samawati Collective has convened the Africa Public Interest AI Council (APiAC) to answer these gaps. The council is deliberately small. Its six members come from journalism, governance, open technology, language AI, data infrastructure and fintech:
– Mr. Kofi Yeboah
– Mr. John-Allan Namu
– Ms. Maha Jouini
– Chido Dzinotyiwei
– Ms. Charlotte N’Guessan
– Mr. Obinna Okerekeocha
They will meet quarterly during an initial two-year term. APiAC is neither a regulator nor an implementer. It is an independent conscience, built around one question: Does Africa’s AI transition expand Africa’s general population agency, democratic participation, cultural expression, and shared prosperity?
The council will commission research, convene public discussions, issue policy advisories to governments and other councils, and offer public input on where philanthropic investment should go.
“We kept seeing the same pattern in rooms where AI strategy was being decided,” said Maurice Otieno, Co-Founder and Executive Director of Samawati Collective. “Plenty of expertise on infrastructure, investment, and regulation, and very little on what any of it means for ordinary people. We didn’t want to build another institution to fill that gap. We wanted a small group of independent minds who could say, in public, what others in those rooms can’t or won’t.”
“The infrastructure gap gets all the headlines, but a gap in who decides is harder to close,” said Kofi Yeboah, Program Officer at the Mozilla Foundation. “Openness and accountability can’t be bolted on after the systems are built. Africa has to help write those norms.”
“My job is to ask who holds power and who answers for it,” said John-Allan Namu, Founder and Journalist at Africa Uncensored Limited. “Those questions now apply to algorithms that shape what voters see. Newsrooms need to be asking them before 2027, not after.”
“Every government is drafting a framework right now,” said Maha Jouini, AI Governance Expert and Founder of Chifaa. “The test of each one is simple: does it protect the person with the least power? Governance isn’t a brake. It’s how a society decides what innovation is for.”
“Language is where the gap is easiest to see,” said Chido Dzinotyiwei, Co-founder and CEO of Vambo AI.”If a model can’t follow how people actually speak, it isn’t serving them, however advanced it is. That has to be a design decision from the start.”
“Data is infrastructure, and the continent is mostly a source of it,” said Charlotte N’Guessan, Lead Data Solutions, AI Ecosystem and Government Partnerships at Amini and Founder, Afrokwary.”Who collects it, who owns it, and who benefits will decide whether Africa takes part in this economy or just supplies it.”
“Millions of Africans already depend on digital systems that decide whether they can save, borrow, or trade,” said Obinna Okerekeocha, Founder, Naija AI Film Festival.”Trust comes from explaining things honestly. AI will earn it or lose it the same way.”
APiAC is maintained independently of funders, governments and commercial entities. Its positions are driven by its members, and it puts public transparency ahead of consensus.
