Africa has its AI strategy, now come the foundations
The continental framework adopted in 2024 hinges on connectivity, electricity and regulators most member states still lack.
Artificial intelligence (AI) is set to accelerate Africa's development, but it also exposes the continent to disinformation, deepfakes, cybercrime and surveillance. The African Union (AU) Continental AI Strategy, adopted by the AU Executive Council in July 2024, seeks to balance harnessing AI's benefits with putting safeguards in place.
The policy process began in 2019, when the AU Specialised Technical Committee on Communication and ICT tasked member states with forming a working group to assess AI's benefits, risks, capability gaps and cooperation needs. The result is a people-centred, development-oriented strategy with 15 action areas across five focus areas, namely harnessing benefits, addressing risks, building continental capabilities, fostering cooperation and stimulating investment.
Leapfrogging is central. AI could allow African economies to skip conventional development stages in agriculture, health, education and public service delivery. The strategy suggests that capturing even a modest share of generative AI's projected global economic value could add over US$100 billion a year to Africa's gross domestic product. This PSC Report examines the pillars, debates and normative commitments that shaped the strategy, and whether it offers a credible pathway to translate this potential into equitable and sustainable development.
A values-driven vision
The strategy moves beyond a technical vision to a values-laden one of what AI should do for Africa. Noting that there is no universal definition of AI, it offers its own, describing AI as computer systems that simulate human intelligence processes, essentially learning, adapting and extracting knowledge from data, to achieve objectives set by humans. This keeps the definition broad and functional rather than tied to specific techniques such as machine learning or generative AI. However, without a clear classification of AI systems, African countries may struggle to gauge potential benefits or harms, their level of exposure and the response required.
AI is framed as something that must be adapted to Africa's context, not the reverse
While AI is consistently presented as a driver of economic growth and social progress, the strategy acknowledges the harms facing a continent entering prematurely into an externally owned and governed environment. These include bias from inequitably sourced data, discrimination against women and vulnerable groups, job displacement, threats to indigenous knowledge, disinformation, surveillance and widening divides between and within countries. The strategy doesn't treat AI as a neutral technology to be adopted uncritically, but as one requiring active governance and shaping.
It insists on an Africa-centric, development-oriented and inclusive approach, citing ubuntu alongside other African norms and universal principles such as human rights and dignity, gender equality and the protection of children and vulnerable persons. AI is framed as something that must be adapted to Africa's context, not the reverse.
From continental ambition to national action
The strategy sets a common direction for 55 member states starting from very different positions. Countries vary considerably in digital infrastructure, technical expertise, research capacity, regulatory institutions and AI talent. The strategy recognises this. Its first phase focuses on governance, national strategies, resource mobilisation and capacity building, before implementation expands through to 2030. The challenge is turning continental ambition into national action.
Some countries are already moving. Rwanda, one of the continent's clearest early examples, approved its national AI policy in 2023, and Ghana developed a national AI strategy for 2023-33. Zimbabwe's 2026-30 strategy emphasises building domestic AI capacity and engaging the diaspora. Nigeria, Kenya, South Africa, Egypt and Ethiopia are pursuing their own approaches. The AU's role is therefore less about imposing uniformity than providing a framework within which countries can develop their own strategies while building continental coherence.
The measure of a successful partnership should be more than the investment it attracts
Non-state actors are also supporting implementation. Smart Africa and development partners such as the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) have provided technical assistance, policy support, training and peer-learning opportunities, including through GIZ's FAIR Forward programme. The programme has supported national AI strategies in several African countries, reached more than 100 000 learners and expanded access to African training data. GIZ estimates that only around 2% of global AI training data originates in Africa, which underscores the need for datasets that better reflect African languages, economies and societies.
The private sector will be equally important. Governments alone cannot build the computing infrastructure, cloud capacity, connectivity and specialised skills AI requires, making public-private partnerships essential. However, the measure of a successful partnership should be more than the investment it attracts. What matters more is whether it leaves behind stronger institutions, skills, infrastructure, data and technical capacity, or deeper dependence on external actors.
Data ownership and control also pose serious sovereignty challenges that African countries must address. For the strategy to deliver, partnerships must strengthen Africa's ability not only to adopt AI, but to shape and govern it.
Where strategy meets infrastructure
Connectivity is the most obvious starting point. According to the latest International Telecommunication Union data, only 38% of Africa's population uses the internet, compared with a global average of 68%. The urban-rural divide is even sharper, with 57% of urban Africans online compared with just 23% in rural areas.
Some countries may formulate AI rules faster than they can develop enforcement institutions
Fixed broadband remains out of reach for most of the population, leaving mobile data as the main route online, and affordability remains a constraint. Only around 1% of Africans have access to 5G. These figures determine who can use AI-enabled services, participate in digital education and work, and ultimately benefit from the technologies governments are promoting.
Electricity is another critical gap. Sub-Saharan Africa accounted for 86% of the world's electricity access deficit in 2024, with 563 million people still living without power. The scale of initiatives such as the World Bank and African Development Bank's Mission 300, which aims to connect 300 million people to electricity by 2030, shows how much basic infrastructure investment is still needed. For countries struggling to provide reliable electricity and affordable broadband, building sophisticated AI ecosystems will be difficult.
Data governance presents a more mixed picture. Forty-four African countries have adopted data protection legislation, a significant increase from a decade ago. But laws alone don't create effective data governance. Only 38 countries have operational data protection authorities, leaving a gap between legislation and the institutional capacity to enforce it. The continent has made considerable progress in putting rules on paper, but less in building the institutions, technical expertise and resources to implement them.
That gap matters for AI because the technical demands are considerably higher. Drafting an AI law or ethics framework is one thing; assessing whether an AI system complies with it requires regulators who understand model behaviour, risk assessment, evaluation methods and the technologies being deployed. Such expertise is scarcer and costlier to build than legislative drafting skills. Some countries may formulate AI rules faster than they can develop enforcement institutions.
This is why regulatory and technical capacity should not be an afterthought to infrastructure and skills. Africa's challenge is not simply to connect more people or adopt more AI systems, but to build the institutions and enabling environments needed to govern technologies that increasingly shape African societies and economies.
The AU Continental AI Strategy is the product of multiyear continental deliberation. It positions Africa not merely as a recipient of AI technology but as a potential shaper of its global trajectory. Its promise rests on leapfrogging conventional development stages and capturing a meaningful share of AI's projected economic value, even as it acknowledges the infrastructure and capability gaps that could widen the continent's AI divide. The strategy provides a concrete blueprint, but its success will depend on how far member states implement and adapt it.
