Benjamin Rosman, one of South Africa’s leading artificial intelligence (AI) experts and director of the WITS Mind Institute, says the country must urgently pursue “algorithmic sovereignty” if it wants to compete globally and ensure AI technologies work for local needs rather than foreign interests.
Rosman, recently named to TIME’s 100 AI 2025 list alongside Sam Altman of OpenAI and Elon Musk of xAI, argues that South Africa cannot simply import AI systems built abroad and expect them to reflect the country’s values, laws, and realities. Instead, it must develop the technical skills, research capacity, and infrastructure required to design, deploy, and govern AI locally.
“Algorithmic sovereignty is the capacity to design, develop, deploy, and govern AI systems locally so that they serve domestic needs, laws, and values,” Rosman writes. “Without these capabilities, we risk becoming little more than exporters of raw digital resources — dependent on foreign companies to process our data and sell insights back to us at high costs, with poor local fit.”
AI, Data, and Digital Dependency
Rosman warns that South Africa risks becoming a “data mine” for global tech giants like OpenAI, Google, and Meta. These companies rely on vast datasets — including African data — to train their AI models but return little value to local economies.
He likens the situation to an oil-rich nation without refineries:
“Just as a nation with abundant oil reserves but no refining capacity remains dependent, a nation with vast stores of data but no local algorithmic capacity is equally vulnerable.”
The dominance of foreign-built AI models, including ChatGPT, Google Gemini, and Meta’s LLaMA, presents multiple risks:
- Cultural Mismatches — AI trained on foreign datasets often misrepresents or ignores African languages, contexts, and social realities.
- Economic Leakage — Valuable African data powers models whose profits largely flow overseas.
- Geopolitical Vulnerability — A sudden withdrawal of foreign AI services due to global tensions could cripple local healthcare, finance, governance, and security systems.
“Relying on foreign AI products creates serious economic and geopolitical vulnerabilities,” Rosman warns. “We must invest in building our own AI ‘refineries’ before it’s too late.”
Decolonizing AI and Building Local Capacity
Rosman calls for urgent action to “decolonize AI” by building African-led systems designed for local realities. This involves:
- Foundational Research — Investing in long-term funding for universities and research hubs to develop homegrown technologies.
- Infrastructure Development — Establishing high-performance computing facilities to process data locally.
- Talent Retention — Expanding competitive research opportunities, mentorship programs, and innovation grants to prevent brain drain.
- Collaborative Innovation — Encouraging partnerships between startups, universities, and government agencies to translate research into viable products and services.
Rosman highlights Deep Learning Indaba, a pan-African AI community, as a model for cultivating and retaining local expertise while promoting cross-border collaboration.
South Africa’s Role in Global AI Governance
Beyond technical capacity, Rosman urges South Africa to actively participate in shaping international AI governance. Rather than adopting frameworks designed elsewhere, he argues the country must push for ethical standards and policies that reflect African priorities and realities.
“Algorithmic sovereignty is not merely about technology,” he writes. “It’s about our right to shape our digital future — refining our most valuable resources locally, ensuring the wealth generated by AI stays within African economies, and aligning technology with our cultural and ethical values.”
A Call to Action
Rosman envisions a future where South Africa becomes a leader in AI innovation rather than a passive consumer of foreign technologies. Achieving this requires:
- Massive investment in local AI infrastructure
- Building world-class research institutions
- Retaining and empowering African AI talent
- Actively contributing to global AI policy debates
“Without decisive action,” Rosman warns, “decisions crucial to our economy, governance, and everyday lives will increasingly be made in boardrooms thousands of kilometers away — whether in Silicon Valley, Europe, or China.”
