In 2024, the African Union adopted its Continental AI Strategy, a policy framework that sets the tone for how African countries approach artificial intelligence.

The strategy rests on five core pillars: applying AI across key sectors such as agriculture, healthcare, and education; building governance systems rooted in ethics and human rights; developing infrastructure, talent, data systems, and research capacity; attracting public and private investment; and encouraging cross-border collaboration.

As artificial intelligence becomes more embedded in everyday life, African countries are racing to design and adopt national strategies to govern and manage AI. Over 15 African countries have launched or published official national AI strategies or policies.

Ghana is one of the African countries that moved early, publishing a National Artificial Intelligence Strategy covering 2023–2033. But nearly two years into its rollout, Ghana is still struggling with implementation and governance. 
South Africa’s approach largely mirrors the AU’s priorities. But instead of translating them into a national law, the country has internalised them into existing systems.

As Nigeria looks to draft its own AI strategy, it needs to analyse the strategies adopted by its fellow African countries and learn from where they have fallen short, especially given that the country boasts the largest population in Africa.

Ghana’s AI strategy was developed through stakeholder consultations with support from international partners such as GIZ’s FAIR Forward initiative, Smart Africa, and The Future Society.
The strategy outlines eight pillars, which are: expanding AI education and training, empowering youth for AI jobs, deepening digital infrastructure and inclusion, facilitating data access and governance, coordinating a national AI ecosystem, accelerating AI adoption in key sectors, investing in applied AI research, and promoting AI adoption in the public sector

The government also identified priority sectors where AI could deliver economic and social benefits, including healthcare, agriculture, transportation, financial services, energy and smart electricity grids, environmental management, and land and natural resource management.
The long-term goal is to position Ghana as a regional hub for AI development and deployment, while ensuring that systems are used responsibly and ethically.

Some parts of Ghana’s AI strategy are already underway. For example, in 2025, the government officially unveiled the strategy at the ENJOY AI African Open event in Accra and announced plans for an Emerging Technologies Bill that would regulate AI, blockchain, and robotics.

The launch of a One Million Coders Programme aimed at building digital and AI skills, the expansion of the government’s technology mandate to include digital technologies and innovation, and collaboration with international partners such as UNESCO and the UK government on AI governance frameworks followed.

Talent development is one visible area of progress. Coding academies, AI bootcamps, university programmes, and initiatives like Ghana Tech Lab incubators are actively training developers and entrepreneurs to build AI products. These programmes aim to address the shortage of skilled professionals capable of building and deploying AI systems locally.

AI governance is also a central pillar of Ghana’s strategy. The policy emphasises responsible AI use, data protection, transparency, and ethical deployment across sectors. To oversee this, the strategy proposed establishing a Responsible AI Office, a body meant to coordinate policy implementation, evaluate AI deployments, and ensure compliance with ethical standards.

“Where the progress is uneven is in the distance between those announcements and operational reality,” Ghanaian AI expert and Founder of the African AI Governance Index, Kwame A. Opoku, says.

While Ghana already has a Data Protection Commission, the dedicated AI governance institutions described in the strategy, including review mechanisms for algorithmic systems used in public services, are not yet operational. 

AI systems are already being deployed in sectors such as financial services, hiring tools, and healthcare. Yet there is no fully operational framework determining how these systems should be audited or regulated.

“Ghana still contends with power reliability issues and limited local compute. The gap is enforcement infrastructure, the specific AI governance architecture that has the legal standing to pull a system when it causes harm is not yet operational,” Opoku explains.

Artificial intelligence requires massive computational resources, a stable power supply, and large-scale data infrastructure, all of which many African countries lack.

Africa currently accounts for less than 1% of the world’s data centre capacity despite representing about 18% of the global population.

Even among AI professionals, access to computing resources is limited. A UNDP analysis of the African AI talent network Zindi found that only 5% of African AI talent have access to sufficient compute power, just 1% has on-premise GPU infrastructure, and most rely on cloud credits or venture funding to run models.

This shortage slows research and development. While developers in advanced economies can train models repeatedly within hours, African developers may wait several days between training cycles.

Although global companies such as Google have established research presence in Accra, a single corporate investment does not yet translate into national AI infrastructure.

Before generative AI tools like ChatGPT went mainstream, South Africa had already begun laying the groundwork.

In 2019, President Cyril Ramaphosa established the Presidential Commission on the Fourth Industrial Revolution (PC4IR). The commission brought together stakeholders across government, academia, and the private sector to shape digital policy, including AI.

Its recommendations helped define South Africa’s AI direction around skills and human capacity development, ethical, inclusive, and human-centred AI, infrastructure and research ecosystems, cultural preservation, and responsible governance.

This gave South Africa a head start, even if policy execution has been slower.

“If you are looking at it from having an AI policy, one could describe it as having been slow. But if you’re also looking at it from understanding the context, one would describe it as being measured,” George Luanda, a South African AI policy expert, says.

Currently, South Africa is finalising its National AI Policy. As of March 2026, the draft policy is moving through Cabinet for approval and is expected to be gazetted for public comment immediately thereafter. Instead of an AI Act, governance is being embedded within existing bodies such as the Information Regulator, the FSCA for finance, and health regulators.

The most tangible progress has been the decentralisation of research through the Artificial Intelligence Institute of South Africa (AIISA). Four major hubs are now operational, each with a specific economic focus.

However, several critical components are missing or under-resourced. While policies exist, AI-specific laws are still absent. The Protection of Personal Information Act (POPIA) remains the primary legal tool for managing AI risks, particularly regarding automated decision-making and data privacy.

Without a statutory framework, there is no clear legal mechanism for redress when an AI system causes harm, leaving victims in a legal grey area, as the POPIA was not designed for training data, deepfakes, or algorithmic bias.

Most national strategies focus on similar pillars: skills development, infrastructure, governance, ethical AI, and public sector adoption. These similarities are partly because many strategies were developed with support from the same international partners and frameworks, like the African Union.

But what ultimately determines success is not the strategy itself; it is the institutions that implement it.

“Nigeria has a population that is bigger than South Africa. The way you would approach the risks is bound to be slightly different, but the historical context will also affect how you do it,” Luanda says. “My concern with Nigeria’s approach would be the capacity of that legislator, particularly the human skills and the financial skills.”

For an AI strategy to work, governance structures must first be operational before AI systems scale widely. Waiting for regulation while AI deployment accelerates can create difficult policy gaps later. In many cases, AI strategies work best when they build on existing digital transformation efforts rather than starting from scratch.

“The specific lesson from Ghana is to not let the Emerging Technologies Bill move at the pace of consensus-building while deployment moves at the pace of the market. From Rwanda, the lesson is legislation. From Kenya, the lesson is letting private sector ecosystem formation precede government prescription,” Opoku says.

Infrastructure and research investment are also just as important as skills training. Bootcamps and digital literacy programmes can produce developers, but without compute infrastructure and research funding, local innovation will struggle to compete globally.

“The most important thing is investing in the youth,” Luanda notes. “Is our education system, which relies on imparting technical expertise (which any AI can now do), still relevant? Google is now offering Nigerian languages natively online. How have they been able to do this without Nigerians benefiting from it? Are the Hausa-speaking people, whose languages and forms are being used, being transformed? Is the only form of transformation going to be them becoming consumers?

AI strategies must also translate into institutions with real authority, budgets, and technical capacity. Otherwise, they risk becoming well-written policy documents that the ecosystem never fully executes.

While AI publications from African universities have grown in recent years, the overall research output remains small. 

Between 2013 and 2022, South Africa produced over 500 AI research publications, and Nigeria published fewer than 100. Most African countries, including Ghana, contribute only a small share to global AI research output. This means research output is growing, but it remains concentrated.

According to Luanda and Opoku, Nigeria must do more than propose an oversight body; it must clearly define its funding model and legal mandate from day one. Without financial and technical independence, any AI Council risks becoming a body that cannot hold large tech actors or government agencies accountable to ethical standards.