AI policy in Africa is shaped not by capacity, but by necessity. Across 99 African Union (AU) policy documents and public statements from 2017 to 2024, AI emerges not merely as a source of innovation but as a driving force for political vision. The AU projects a confident, future-oriented AI strategy - one that uses policy to navigate constraints, mobilise institutional will, and assert an African way to AI before the material conditions exist to sustain it. However, amid acute infrastructure and resource gaps, questions remain about how achievable these aspirations will be, writes Aleksei Turobov.
Can you lead a digital revolution when nearly 600 million people lack access to electricity? Can you legislate for autonomous weapons while non-state actors already deploy drones across conflict zones? Can you talk of sovereign artificial intelligence (AI) ethics when fibre connectivity still eludes 300 million citizens? These are not rhetorical questions but the reality in which the African Union operates.
Over seven years the African Union (AU), consisting of 55 member countries across the continent, has developed one of the world’s most ambitious and versatile AI policy architectures. From youth innovation to media freedom, from sovereign ethics to climate adaptation, from industrialisation to existential threats, the AU’s AI strategy documents are not merely reactive to global developments but rather insist on the scope for a future shaped to African needs. Yet rather than deferring ambition, the AU treats the absence of capacity as the very reason to act, transforming limitations into a lever to develop a strategic vision.
However, apart from the 2024 Continental AI Strategy and a handful of directives, most of the 99 documents are broader digital policies or communiques, signalling non-binding mandates, overlapping drafts, and fragmented AU-level coordination on AI. Although public debate on African AI and technological development often fixes on China – and, to a lesser extent, the US and EU – the corpus and our analysis contains only scattered references (a China’s Belt and Road initiative and occasional parallels against Europe/US approaches), consistent with the AU’s African-led, multi-vector stance that preserves diplomatic room to manoeuvre while advancing a sovereign AI agenda.
As in other regions globally, the policy statements regarding AI have shifted. Since 2017, the African Union’s engagement with AI has evolved from overoptimistic expectations to institutional governance and sectoral implementation. The policy approach began with grand proclamations –“A.I. is considered the last invention of humanity“ – and matured into precise directives – “Develop AI strategies in critical sectors including education, agriculture, health, industry, peace and security“ (2024).
Yet these narratives unfold against harsh reality – “Africa has the lowest internet usage rate in the world, at 37%, compared to a global average of 67%,” and “just over 40% of the continent’s population has access [to electricity]“ (2024). The very foundation required for digital transformation remains elusive, even as AI policy and deployment accelerate. What makes the AU’s approach so instructive is that it does not deny these limitations – it integrates them. The absence of infrastructure is treated as a reason to act. In this sense, the AU positions AI policy as a political instrument to catalyse it, using governance frameworks, ethical declarations, and strategic planning as tools to mobilise resources, build institutions, and shape a future that might otherwise remain out of reach.
In 2017, the AU stated: “The full implementation of Robotics and Artificial General [Intelligence] will allow full integration of all industries and maximum value creation.” By 2019, a pivotal shift occurred, proposing to “establish a working group on Artificial Intelligence to study… the creation of a common African stance on Artificial Intelligence (AI).” In parallel, AI policy became grounded in sectoral applications. In agriculture, for example, the AU highlighted how “Promagric helps farmers combat crop diseases to reduce agricultural losses… using artificial intelligence to diagnose crop diseases from an image“ (2021). In education: “We used AI and design thinking to lessen the ed-tech learning curve that teachers have to endure in the 4IR“ (2021). By 2020, even COVID-19 was reframed not merely as a crisis, but as a catalyst: “The crisis of COVID-19 has become the biggest catalyst for large scale digital transformation in Africa.” What begins as ‘myth’ – using terminology such as the last invention, the final threshold of intelligence – evolves rapidly into a focus on practical mechanisms.
While many regions treat entrepreneurship as a catalyst for innovation, the African Union frames it as a structural response to institutional inadequacy. That is, where weak state capacity limits public job creation, infrastructure investment, and R&D, entrepreneurship is positioned not as optional innovation, but as essential substitution. From the earliest years, startups were positioned as the vanguard of technological development: “From energy and healthcare to agriculture and financial services, African startups are bringing innovative ideas and business models that need support“ (2017). This emphasis intensified over time such that by 2019 the AU-backed A-eTrade Group would “contribute… to the creation of 600,000 SMEs in Africa that will generate 22 million jobs,“ recognising a limitation: “The state cannot create the number of jobs needed for the large population that enters the labor market every year.“
By 2024, this entrepreneurship narrative matured into an explicit AI focus: “Accelerate the adoption of AI by the private sector, including small and medium-sized enterprises, and create an enabling environment for AI startup ecosystems focused on solving Africa’s development problems.” But the policy documents also recognise that structural gaps persist: “Only 5.4% of the total funds raised went to startups younger than five years old” (2021) and “Limited technical know-how and weak processing capabilities stifle production” (2022). Entrepreneurship, in this narrative, is a workaround given the African context. Where the state cannot deliver, startups must. Even so, the stated policy ambition still outpaces capital, skills, and ecosystem maturity.
Across nearly a decade of policy documents, the African Union consistently frames the continent’s young people as the principal agents of AI transformation. In 2018, for instance, the AU called youth “Africa’s most precious asset“ and committed to remodelling, harnessing and empowering this majority demographic. This rhetorical baseline is rooted in demographic urgency: “In Africa, while only about 3.1 million jobs are created each year, 10 to 12 million young people are in search of employment“ (2020).
The AI narrative builds on this urgency, describing youth as critical to continental technological sovereignty: “By investing in African youth, innovators, computer scientists, data experts, and AI researchers, the framework paves the way for Africa’s success in the global AI arena.“ (2024). Yet in this respect too, the policy documents acknowledge a profound gap between aspiration and capacity. In 2024, the AU admits: “Three quarters of Africa’s youth [are] without digital skills.“ And in 2023: “One-third of these young people are unemployed and discouraged.” The consistent emphasis on young Africans is powerful and politically strategic. But without sustained investment in education as well as physical connectivity, the demographic dividend may never convert into digital leadership.
The African Union frames AI governance as a civilisational project. Its documents consistently reject imported ethics and regulatory ‘mimicry’ in favour of homegrown values-based frameworks. In 2018, the AU reaffirmed this by proposing to “develop a Pan-African Charter on Ethical AI,“ a foundational governance instrument rooted in African contexts and priorities. In 2022, a cybersecurity expert declaration demanded: “As Africans, we need to articulate our own Philosophy, Ethics, Policy, Strategies and accountability frameworks for Cyberspace, Cybersecurity and Cognitive or Artificial Intelligence (AI).” Importantly, this drive for sovereignty is also strategic. In 2024, the AU called for ensuring that “governance mechanisms for Artificial Intelligence (AI) reflect the perspectives of the Global South.” AI ethics is not framed as a neutral technical discipline, but as a geopolitical arena where voice, history, and justice must be reclaimed. In a world where algorithmic systems often reinforce historical hierarchies, the AU insists that ethics must be contextual, accountable, and African by design.
The 2024 Continental Artificial Intelligence Strategy offers a clear technical definition: “AI refers to computer systems that can simulate the processes of natural intelligence exhibited by humans where machines use technologies that enable them to learn and adapt, sense and interact, predict and recommend, reason and plan.” Here, the language is algorithmically literate, methodologically defined, and sectorally anchored. AI is no longer just a future disruptor – it is a policy object. The Continental Strategy also operationalises how the African Union intends to use it, govern it, and embed it across education, agriculture, healthcare, industrialisation, peace, and security.
The African Union’s AI narrative takes a darker turn in its later years, as the documents begin to confront the weaponisation of artificial intelligence and the erosion of state control in conflict zones. In 2024, the AU acknowledged a critical escalation: “Such systems [autonomous weapons] have been easily acquired by non-state actors, with disturbing consequences, particularly across the continent.“ The concern is not hypothetical – “In Libya, warring parties, with support from external actors, routinely employ autonomous weapons systems as force multipliers.” And also, “Al-Shabaab has leveraged knowledge transfer from other conflict zones to utilise drones for reconnaissance and surveillance of African Union troops.”
This shift reframes AI as a dual-use technology – capable of both developmental transformation and destabilisation. It also introduces a new regulatory imperative. The AU has now called for a “legally binding agreement on autonomous weapons systems,“ a position that contrasts sharply with more hesitant international stances. While early AU narratives celebrated AI as the key to leapfrogging development, the later years warn of leapfrogging violence, where advanced technologies bypass weak institutions and enter conflict theatres directly.
One of the most striking – and globally underappreciated – aspects of African Union AI policy statements is their consistent attention to the democratic risks of AI, particularly in relation to media freedom. Already in 2022, the AU warned of “big data collection and the negative impact of artificial intelligence (AI) on the work of the media, which erodes the rights to freedom of expression and opinion.” This concern deepened: “Journalists are often the soft targets of surveillance, as big data collection and the negative impact of artificial intelligence (AI) on the work of the media.”
By 2023, the analysis had expanded to include economic threats: “Draw attention to media viability and sustainability of legacy media given rising concerns about their survival… and the advent of Artificial Intelligence Generative Pre-Trained Transformer.“ And by 2024, an AU statement directly linked AI to information warfare: “risks that these technologies pose to media freedom and freedom of expression including as tools for coordinated and targeted disinformation campaigns.”
Where many regions’ policy statements portray AI through the lens of innovation or industrial competitiveness, the AU positions it as a threat to democratic infrastructure. This makes Africa not a latecomer to AI governance debates, but a potential global leader in norm-setting around digital rights and press freedom.
However, beneath the ambition of the African Union’s AI strategy lies a sobering and persistent reality: the continent’s infrastructural base is deeply inadequate for the technological future it envisions. In 2024, the AU reports: “About 590 million people in Africa lack access to electricity.“ In a 2020 report, the AU notes: “Nearly 300 million Africans live more than 50 km from a fibre or cable broadband connection.” Despite these figures, the vision remains expansive: “By 2030 all our people should be digitally empowered and able to access safely and securely to at least (6 mb/s) all the time.” (2020) This (miss)match creates a deep tension, or what may be better understood as a strategic paradox. The AU documents never deny these limitations. Instead, they make the absence of infrastructure part of the rationale for immediate, ambitious action. In this context, AI policy becomes not the outcome of development, but the instrument through which development is imagined, demanded, and politically realised.
Across AI, space, digital finance, climate, and industrialisation, AU documents explicitly tie emerging technologies to decades-long into the future. In 2019, the AU outlined the African Space Strategy, focusing on “Earth observation, Satellite communication, Navigation and positioning, and Space Sciences and Astronomy.” It projected AI’s economic value: “PwC estimates that AI could contribute up to $1.5 trillion to the African economy or 6% of the continent’s GDP.” At the same time, the documents remain candid about present limitations: “Insufficient regulations, a lack of bankable projects, difficult financing conditions and poor energy infrastructures represent major structural barriers“ (2023). Rather than treat these as contradictions, the AU performs its future as deliberate narrative coordination – using long-range planning to stabilise political will and express sovereign agency.
The African Union’s AI policy architecture is forged amid scarcity, fragmentation, and historical asymmetry. And yet, rather than mimic the regulatory templates of AI powers or retreat behind capacity constraints, the AU insists on governing the future as though it is already within reach. Where most global AI strategies are reactive – focused on regulation, acceleration, or containment – the African Union’s vision is active and generational. The backbone of this orientation is Agenda 2063, the AU’s long-term blueprint for continental transformation. It projects almost five decades forward – an unusual horizon in an era marked by short-term crises and volatile technological cycles.
This is not naive optimism – it is a deliberate strategy of ambition under constraint. Across the 99 documents analysed, the AU does not wait for capacity to precede vision; it uses vision to provoke capacity into being. But therein lies the paradox. While the narrative is compelling, turning constraint into momentum and policy into an instrument of future-making, it risks detaching vision from institutional viability. It is one thing to declare sovereign AI ethics (an African approach to AI) or industrial innovation zones; it is another to build institutions that can enforce, fund, and sustain them. The decisive test is the near-term targets outlined in the 2024 AI Strategy, which are to be delivered by 2025. Progress here will indicate whether coordination can transition from a narrative to a practical application.
In many ways, the AU’s AI strategy uses policy to shape futures that remain materially out of reach. But it also surfaces a deeper question for Africa and the world: Can ambition alone build infrastructure? Can narrative alone create institutions? That is the real lesson from the AU: ambition without infrastructure, vision without mechanisms, frameworks without enforcement, can collapse into confusion – when policy as a narrative drifts too far from policy as implementation.
This AU analysis is part of the AIxGEO project, which examines international approaches to AI governance, and is the third blog following APEC and ASEAN. Our previous analysis explored the policy debate in Western institutions – Moving beyond competition: domain-specific approach for international AI framework. Forthcoming analyses of the G20 initiatives will provide further insights into how diverse multilateral forums develop distinctive governance models adapted to their regional contexts, offering a more nuanced understanding than Western-centric debates of how effective governance emerges across different geopolitical environments.
Image: Flag of the African Union, Wikimedia
The views and opinions expressed in this post are those of the author(s) and not necessarily those of the Bennett Institute for Public Policy.
