The Global South Is Rethinking AI But Not Yet Controlling It
From Johannesburg to New Delhi, governments are reframing artificial intelligence around justice and sovereignty. But control over infrastructure and governance remains concentrated elsewhere.
A set of questions is becoming central to global debates on artificial intelligence: who builds the systems, who governs them, and whose interests they reflect?
At the Just AI Conference in Johannesburg this week, policymakers and researchers gathered for discussions that began with ethics and bias but quickly moved to something more structural: control.
“The question is not whether AI can govern our infrastructure, but whose justice will be entrenched into its architecture,” said Lavina Ramkissoon, an adviser to the African Union on science and technology. “Efficiency without equity is merely the automation of exclusion.”
The concern is shared across much of the Global South. But while the framing is shifting, control over AI systems remains concentrated elsewhere.
Participation without control
For policymakers within the A.U. system, the challenge is not only how to regulate AI but how to reposition the continent within the systems that make it possible.
“Africa largely participates at a consumption layer,” said Pamla Gopaul, who leads economic analysis and foresight at the A.U. Development Agency and heads the Africa Policy Bridge Tank. “We don’t participate at an infrastructure or governance layer.”
The imbalance is stark. Despite accounting for roughly a fifth of the world’s population, Africa hosts less than 2% of global data-center capacity, the physical backbone of AI. Much of that infrastructure is owned or financed by external companies, reinforcing dependence not only on technology but on the systems that sustain it.
That creates a clear imbalance: growing regulatory ambition alongside continued dependence on systems built and controlled elsewhere. “There is a risk,” Gopaul said, “that Africa becomes a well-regulated system within designs made elsewhere.”
In India, the response has been to build. A recent AI summit drew 250,000 registered attendees and saw more than $200 billion in commitments tied to data centers, computing capacity and digital infrastructure. Prime Minister Narendra Modi framed the country not as a user of AI systems, but as a builder of them.
The shift is from participation to production, but it remains uneven.
Competing systems
In Brussels, artificial intelligence is increasingly framed as a geopolitical contest. The United States leads on innovation speed, China on state-backed scale, and Europe on regulation and standards.
Other approaches are taking shape in Asia. Taiwan, for example, is pursuing what it describes as “sovereign AI” — combining domestic investment in infrastructure with integration into technology ecosystems aligned with the United States and Europe.
The Global South is often positioned as a market. That risks reproducing older patterns of dependency, said Baratang Miya, a United Nations consultant on digital public infrastructure and founder of GirlHype Coders Academy, a South African nonprofit that provides coding and digital literacy training to women and girls.
She warned that Africa’s role in global technology systems has long been externally defined. Whether treated as a market or a consumer, she argued, the continent risks reproducing a “perpetual history of colonization” in AI.
At the same time, she pointed to a quieter shift: African researchers and developers are modifying imported systems, including adapting language models to African languages and local data contexts, embedding knowledge into models not originally designed for their societies.