The Algorithm#1
AI across the Global South, tracked weekly
AI is being built, regulated, and resisted across the Global South at a pace that rarely makes it into mainstream tech coverage. Each week, The Algorithm tracks the most important developments : new models, policy moves, funding gaps, and the harder questions about who AI is actually for.
India puts itself at the center of global AI governance
Press Information Bureau, India · March 11, 2026
At its AI Impact Summit in New Delhi, India secured over $250 billion in investment commitments from 92 countries and got 13 major model providers to sign the “New Delhi Frontier AI Impact Commitments.”
More notably, India released practical AI casebooks for agriculture and healthcare built specifically for Global South contexts. This is a direct bid to be a rule-shaper, not a rule-taker, in how AI gets built and deployed globally.
The language wall is getting chipped away , from two directions
Tech-Ish (Kenya) · March 5, 2026 / UCT News (South Africa) · March 18, 2026
The GSMA launched the world’s first open Swahili reasoning model at MWC 2026, built with Zambian and Zimbabwean partners and designed to run on $40 smartphones.
Separately, a consortium of four South African universities, backed by the National Research Foundation and Telkom, announced a national effort to build LLMs for isiXhosa, isiZulu, and Sepedi, aimed at improving healthcare in communities that have never had a model that speaks their language.
Two very different approaches, one commercial, one institutional, converging on the same problem.
Nigeria: 122 startups, 79% can’t raise
Leadership Nigeria, citing Nigeria AI Outlook 2026 · March 2026
The Nigeria AI Outlook 2026 mapped 122 active AI startups, including Ubenwa Health, which diagnoses infant neurological conditions by analysing a baby’s cry, but found that nearly 4 in 5 founders struggle to access capital, with 40% bootstrapping entirely. The ecosystem is real, however, the money is not.
The report outlined that this funding gap shows that Nigeria’s AI innovation economy is still fragile. “Without strong local investment networks or government support, many founders are forced to grow their companies with limited resources, often putting long-term survival at risk.”
Kenya wants to send you to jail for deploying unapproved AI
TechCabal · March 17, 2026
Kenya’s proposed Artificial Intelligence Bill 2026 would criminalise the deployment of unapproved high-risk AI systems in finance, health, and biometrics — with company directors facing personal liability. If passed, it would make Kenya one of the first African countries with dedicated AI legislation that carries criminal enforcement.
Brazil draws a hard line between AI and elections
Courthouse News Service · March 5, 2026
Ahead of its 2026 general election, Brazil’s Superior Electoral Court issued 14 new resolutions banning AI chatbots from offering voting advice, requiring mandatory labelling for AI-generated campaign content, and imposing an AI media blackout around election day. Brazil has quietly become one of the most assertive regulators of AI’s role in democracy, without waiting for a national AI law to pass first.
Uruguay and Microsoft open South America’s first AI for Good Lab
BNAmericas · March 16, 2026
The Government of Uruguay has partnered with Microsoft to launch South America’s first AI for Good Lab, a facility focused on health, education, and government services. It is a small country making an outsized bet and the model is notable: neither the US commercial approach nor the EU’s regulation-first posture, but a state-directed, public-good-oriented third way that other Latin American governments will be watching closely.
Western AI doesn’t know what a cassava plant looks like
Rest of World · March 12, 2026
Agricultural AI models built on Western training data routinely fail to identify local crops in Kenya and forest types in India — producing outputs that can actively mislead the farmers relying on them. As the global digital farming market approaches $84 billion, organisations like Digital Green and NASA Harvest are building localised datasets as a counter. But the structural bias in how training data gets assembled is a design choice, not an accident.
That’s it for this edition of The Algorithm. If you spotted a story we missed — a startup, a policy move, a community using AI in ways nobody’s writing about yet, reply to this email. The best leads come from readers on the ground.