By 
By Dr Josefin Rosén (SAS Institute), Dr Rachel Adams and Selamawit Engida Abdella (Global Center on AI Governance)
New report shows how the Global South can drive the development of ethical and responsible AI.
A new report from the AI company SAS Institute and the Global Center on AI Governance challenges the image of the Global South as a region that is lagging behind in AI development. Instead, the report highlights crucial opportunities for these countries to take an active role in shaping the AI of the future, with investments in capacity building, digital infrastructure and inclusive governance.
RELATED: Africa’s AI skills landscape revealed in new SAP report
The question of how to avoid a growing global AI divide has taken up a lot of space on the international agenda in recent weeks, not least during the World Economic Forum in Davos.
The report, Constraint to Capability: Flipping the Narrative on AI in the Global South, is authored by Dr Josefin Rosén, Principal Trustworthy AI Specialist at SAS Institute, together with Dr Rachel Adams and Selamawit Engida Abdella from the Global Center on AI Governance.
From Constraint to Catalyst: Reframing Limitations as Strategic Opportunities
It argues that structural constraints such as limited infrastructure, data gaps, and governance immaturity can be wielded into become strategic advantages if addressed deliberately. For African markets, including South Africa, this carries practical implications for economic competitiveness, digital sovereignty, and long-term stability.
The global AI economy is accelerating. But much of today’s AI development remains shaped by Western-centric datasets, infrastructure, and governance models. That concentration risks embedding new forms of dependency and limiting emerging economies’ ability to shape AI systems that reflect their languages, social realities, and regulatory priorities.
“The global AI gap risks creating a new form of inequality. But if we direct investments to the right areas, such as AI skills, access to representative data, and inclusive governance models, the Global South can play an active and meaningful role in shaping the future of AI. It’s about shifting the perspective, from limitations to opportunities,” says Dr. Josefin Rosén, Principal Trustworthy AI Specialist at SAS Institute.
A demographic and digital advantage
The report highlights two structural dynamics that favour the Global South. The first is demographic. Most of the world’s young and digitally native population resides in these regions. Across Africa, digital adoption has leapfrogged traditional infrastructure, driving growth in mobile banking and other digital services.
South Africa, with its comparatively advanced financial services sector, established universities, and growing technology ecosystem, sits at a strategic intersection. It has both the institutional capacity and the regulatory frameworks that could position it as a regional anchor for the responsible development of AI.
However, that potential depends on deliberate investment in AI literacy, research capacity, and digital infrastructure. Without it, the continent risks becoming a consumer of imported AI systems rather than a contributor to their design.
Governance as a competitive advantage
A second advantage identified in the report is institutional flexibility. Countries that are not burdened by decades of legacy AI systems may have greater freedom to design governance frameworks from first principles.
For African policymakers and technology leaders, this means the opportunity to embed ethical safeguards, transparency requirements, and data sovereignty protections from the outset, rather than retrofitting them later. Africa’s AI trajectory will be shaped by how quickly institutions translate policy ambition into scalable operational governance frameworks and skills programmes.
“AI sovereignty is not about isolation. It is about ensuring that AI systems reflect local contexts, languages, and values, and that countries retain meaningful control over the data and models that influence their economies. For African countries, the choices made today around skills development, infrastructure, and governance will determine whether AI becomes a driver of resilience or a source of new dependencies,” says Selamawit Engida Abdella from the Global Center on AI Governance.
Representative data and inclusion
One of the report’s central concerns is representativeness. AI systems trained predominantly on Western datasets risk producing biased or incomplete outcomes when applied elsewhere.
In practical terms, this can affect everything from credit scoring and healthcare diagnostics to public service delivery. In multilingual and socioeconomically diverse societies such as South Africa, unrepresentative models may entrench exclusion rather than reduce it.
The report, therefore, calls for investment in local data ecosystems, including the development of language models that reflect indigenous knowledge and local languages of the region. It also explores the responsible use of synthetic data as a mechanism to increase representation while protecting privacy and managing data scarcity.
For South Africa, where questions of data protection and sovereignty are already embedded in regulatory frameworks, this creates an operational challenge: balancing innovation with compliance, and scale with accountability.
Infrastructure and capacity building
The report does not understate the structural constraints facing the Global South. Infrastructure gaps, limited compute capacity, and uneven digital access remain real barriers. However, it argues that these constraints can inform more sustainable and context-appropriate AI development models.
Rather than replicating capital-intensive AI architectures designed for advanced economies, African markets can focus on targeted, high-impact use cases in areas such as health, education, and financial inclusion.
Public-private collaboration is highlighted as critical. Universities, regulators, technology firms, and industry bodies must coordinate on skills pipelines and governance frameworks that enable experimentation without compromising public trust.
For South Africa, where policy debates around digital transformation and economic competitiveness are ongoing, this suggests that AI strategy cannot be confined to isolated technology initiatives. It must sit within a broader national development agenda.
From participation to influence
The report ultimately argues that inclusion in the AI economy is not a question of charity, but of global stability. AI systems that exclude large populations from participation or misrepresent their realities create systemic risks.
For African countries, the objective should not be to “catch up” to established AI powers, but to shape AI systems that are locally relevant and globally interoperable.
It entails investing in AI skills and research capacity, building representative datasets, embedding transparent and enforceable governance, and structuring partnerships that strengthen, rather than dilute, sovereignty.
The window for shaping this trajectory is narrow. AI systems are already being deployed at scale across finance, public services, healthcare, and infrastructure. The decisions made now will determine whether African economies become rule-takers or rule-makers in the AI era.
As Rosén notes, “This is about shifting perspective from limitation to capability. With the right investments and policy decisions, countries across Africa can help define what responsible and inclusive AI looks like globally.”
For South Africa, the question is no longer whether AI will reshape the economy. It is whether the country will help shape that transformation or simply import it.
About the report
Constraint to Capability: Flipping the Narrative on AI in the Global South was published in December 2025 and is a collaboration between SAS Institute and the Global Center on AI Governance. The report provides a strategic roadmap for how countries in the Global South can take a stronger role in the global AI economy through investments in skills, infrastructure and inclusive governance.
Read the report:
