Available sources do not allow for a precise determination of how the number of domains covered by African national AI policies has evolved. This regulatory dynamism is real, but the figures do not clarify whether these frameworks are building sovereign technological capacity or organizing the institutional reception of models designed elsewhere. AI governance without local productive mastery can contribute to technological dependence.
The Essential Points
- Between 2024 and 2026, the average number of domains covered by African national AI strategies increased from 2.5 to 4.7, a rise of 83% (Digital Mag Côte d’Ivoire, 2025).
- Côte d’Ivoire and Zambia have launched national AI strategies accompanied by concrete funding: 80 million dollars from the African Development Bank for Abidjan, a target of 8% GDP growth by 2030 for Lusaka.
- These frameworks primarily regulate the use and importation of foreign proprietary models, without symmetrical investment in the local production of models or data infrastructures.
- The structural risk is that of regulatory capture: foreign technological powers contribute to shaping the standards that frame their own products on African markets.
- The bifurcation will play out over the 2026-2035 decade: Africa can still orient its governance toward technological autonomy, but the windows of investment in training, data, and infrastructure are closing quickly.
Strategies That Cover More Ground Without Digging Deeper
The progression is notable on paper. African national AI strategies cover multiple sectors. These sectors already appear in official continental texts prior to or contemporary with 2024; several recent national strategies take them up or clarify them. Zambia launched its strategy in 2024; the official document is the National AI Strategy 2025-2027. Côte d’Ivoire received 80 million dollars from the African Development Bank for the PARAE project; the AI strategy was presented at the launch of this project.
Egypt and Rwanda have advanced their own frameworks with specific sectoral objectives.
This movement is encouraging. It signals that African governments take AI seriously, refuse to leave its definition solely to foreign private operators, and seek to direct its benefits toward national priorities. Regulation is not a brake on technological development: it is often its condition. One does not capture the returns of a technology one has not defined the right to use.
Expanding the number of domains covered says nothing about the nature of what is being regulated. National and continental strategies generally frame the AI ecosystem and its uses; they also include, to varying degrees, objectives for local capacities and innovation. A significant portion of models used in Africa comes from foreign companies, but the frameworks also apply to local systems, open source or developed in other jurisdictions. Regulating the use of these models is legitimate and necessary.
Without investments in local capacities, regulation can be applied to a market dominated by foreign suppliers.
The Gap Between Governance and Productive Capacity
The imbalance is structural. Building a regulatory framework requires lawyers, consultants, policy experts, and several years of institutional work. Building productive capacity in AI requires GPUs, trained engineers, corpora of data in local languages, electrical infrastructure, and long-term policies. The timelines are not the same. Neither are the budgets.
Côte d’Ivoire benefits from 80 million dollars in financing from the AfDB for the PARAE project, a broader public digital program including components relevant to data and AI. The gap in computing power between sub-Saharan Africa and American or Chinese data center clusters will not be closed by national strategies, however well-written they are, if they are not accompanied by coordinated investments in physical infrastructure.
This imbalance is not unique to Africa. As the dynamics of submarine cables and global digital infrastructure show, controlling the physical layers of the network conditions who controls the services circulating on top of it. AI governance obeys similar logic: mastery of foundational models can influence the conditions of use. A regulatory framework without local productive capacities may not modify this distribution of power.
Training is another blind spot. Several African strategies mention developing local AI skills, but university programs capable of producing machine learning researchers remain rare and underfunded. Without this human resource, African companies risk depending more heavily on foreign solutions. And the talent that emerges often leaves, to where salaries and research infrastructure exist.
Regulating Can Mean Validating
There is a risk that African institutional circles discuss little openly: regulatory capture by the actors one claims to oversee. This phenomenon is documented in other sectors, notably finance, telecommunications, and pharmaceuticals.
Major foreign technology actors have every interest in actively participating in the drafting of African regulatory frameworks. This participation can give them influence over the standards applicable to their products. Participation in public consultations varies by country, sector, and organizational modalities.
Geopolitical fragmentation complicates governance choices: the United States and China each promote their own approaches to AI. African governments can draw inspiration from these frameworks while defining their own priorities.
This does not render regulation useless, but it requires clarifying its purpose. Protecting African citizens from abusive uses of AI and securing the conditions of entry for foreign suppliers on African markets are two compatible objectives, which nevertheless do not lead to the same public policy choices.
Strategies That Produce Results
The picture is not uniformly bleak. A few trajectories merit close attention because they attempt precisely to combine governance and productive capacity.
Rwanda has invested in a structured public data infrastructure, digitized civil registries, agricultural data, and health data, which constitutes an asset on which local models can be trained. This foundation is not yet an AI industry, but it represents a prerequisite that most other African countries have not yet met. Quality public data is the raw ore of models: without it, technological sovereignty remains an empty formula.
Egypt seeks to develop its digital infrastructure and regional technological capacities. The strategy is debatable in its details, but it recognizes that governance is not enough, that material capacities are also needed.
At the continental level, the African Union is working on a common framework to avoid regulatory fragmentation among member states, which would benefit chiefly external actors capable of navigating multiple jurisdictions. This coordination is slow, but it is moving in the right direction: the size of the consolidated African market is a negotiating lever that isolated countries do not have.
African private actors are also beginning to produce models trained on local data and in African languages—Swahili, Yoruba, Amharic, Wolof. These initiatives remain fragile and underfunded, but they prove that local AI production is technically possible. They need public procurement, open data, and a regulatory framework that reserves them a place against foreign giants, rather than ignoring them.
The Bifurcation of the Decade: Two Trajectories for 2035
The decade opening is the one where positions are fixed. AI is not a static technology that can be adopted later at lower cost: foundational models are already being trained. If dependence takes hold, it will be difficult to undo.
Two trajectories are emerging, though neither is yet determined.
In the first, national strategies may encounter difficulties in execution or financing despite objectives also covering productive capacities. Foreign models then occupy an important place on African markets, while the development of local capacities depends on investments, skills, and public procurement.
In the second, national strategies can serve as a negotiating lever with donors and technology partners to obtain skills transfers, investments in training, and research partnerships. The AfDB and the World Bank support both governance frameworks and the material and human foundations of AI. Public procurement reserved for locally developed solutions creates demand for African AI startups. This trajectory is more politically demanding, but it is feasible.
What would make it possible to distinguish the two trajectories in the coming years hinges on a few observable signals. The first is the structure of financing: the allocation of funds between governance, infrastructure, training, and public data will make it possible to assess the priorities retained. The second is the composition of public markets: if African administrations purchase exclusively foreign solutions for their own AI needs, they deprive local actors of the initial demand they would need to develop. The third signal is the participation of African startups in regulatory consultations; their absence or marginal presence is a reliable indicator of the direction the frameworks are taking.
In the next two or three years, African governments will need to determine whether their national AI strategies are tools for sovereign technological development or conditions for welcoming technologies developed elsewhere. The answer is not found in the texts, but in the budgets, public procurement, and partnerships that these strategies concretely generate.
Sources
- Digital Mag Côte d’Ivoire, Artificial Intelligence Strategies in Africa, 2025 State of Play: https://digitalmag.ci/strategies-dintelligence-artificielle-ia-en-afrique-ou-en-sont-les-pays-en-2025/
- African Development Bank, Report on Digital Strategy and AI 2025 (AfDB, no verified link available)
- Zambia National Artificial Intelligence Strategy 2024-2027 (Ministry of Technology, Zambia, no stable verified URL)
- African Union, Continental Framework for Artificial Intelligence (AU Commission, no verified link available)