The AU’s Question: What 55 Countries Need to Agree on Before Africa Can Govern AI  

Africa’s debate over how to govern artificial intelligence is increasingly running into a more basic question: what does the continent need to control, share and build together if it wants AI to deliver value locally?

That question ran through the opening day of the African Union’s Open Consultation on AI Governance and Regulation on Wednesday, September 23, 2026, as the AU’s Infrastructure and Energy Department opened a two-day session meant to guide how Africa approaches AI over the coming years. Convened to operationalize the Continental AI Strategy endorsed by the AU Executive Council in Accra in July 2024, it brought together policymakers, technologists, civil society and industry to examine the institutional mechanisms needed to align the continent.

Day one’s panel, AI Governance in Africa, was moderated by Souhila Amazouz, Principal AI & Digital Policy Coordinator at the African Union Commission. It brought together Mehdi Snene of the UN Office for Digital and Emerging Technologies, Dr Girmaw Abebe Tadesse of Microsoft’s AI for Good Lab, Lillian Nalwoga representing civil society, and Angela Wamola of GSMA Africa. Opening the proceedings, Amazouz noted that while national adoption is accelerating, alignment remains critical:

“When we developed the strategy, we had eight countries that formally adopted national AI policies. Now we have 17, one-third of African nations. For Africa, we need to come up with a governance and regulation approach that aligns with the African context, responds to national priorities, and ensures harmonization across the continent.”

– Souhila Amazouz, Principal AI & Digital Policy Coordinator, African Union Commission 

Strategic Autonomy Over Technological Isolation 

Addressing the panel, Angela Wamola, Head of Africa at GSMA, argued that national AI regulations will achieve little if they remain disconnected from physical infrastructure. For the telecommunications and digital industry, sovereignty cannot be pursued through fragmented national projects that ignore the economics of scale.

“Sovereignty for us is not to be seen in technological isolation or an ambition for someone to achieve something independently. Everything, the whole chain from data centres, energy, infrastructure, etc. But I think we should see it more as a strategic autonomy. That’s what we want to achieve as Africa and not nationally, because that’s where we begin to lose things as a continent.”

– Angela Wamola, Head of Africa, GSMA

Wamola cautioned against attempting to impose identical, rigid AI laws across 55 diverse jurisdictions. Instead, she urged the adoption of a continental baseline that allows member states to progress according to their digital maturity while maintaining cross-border compatibility.

“We’re not talking about imposing identical legislation. What we’re talking about is a continental baseline, interoperable rules and standards that will allow every country, depending on where they are on their maturity level of frontier technologies adaptation, to progressively come towards harmonizing all that we are talking about here in terms of AI governance,” she said.

Shea also highlighted data governance, cross-border data flows, cybersecurity, and sandbox testing for high-impact AI systems as priority areas where alignment can prevent market fragmentation and lower entry barriers for local innovators. It’s a framing the rest of the panel would return to from different angles, whether they were talking about UN leverage or testing AI in the field.

The Physical Reality: Data, Compute, and Energy as Shared Assets 

The discussion repeatedly returned to the physical constraints underpinning AI deployment: energy grid reliability, fiber connectivity, and high-performance computing capacity.

Wamola pointed out that individual national ambitions to build standalone data infrastructure in every country risk wasting scarce capital. Instead, computing power and data flows must be treated as integrated continental utilities. She cited consumer adoption trends to illustrate the risk of remaining passive consumers in global supply chains:

“Crucially, this sovereignty and harmonization must have a view of commercial sustainability. Did you know that when we measure internet users active on ChatGPT globally, Kenya ranks number one? Our people are busy feeding data online, but because we are not harmonized continentally, that information leaves without bringing back local economic empowerment. We remain users and not producers. This requires us to treat data, compute, and connectivity as an integrated foundation for Africa.”

-Angela Wamola, Head of Africa, GSMA 

 Leveraging 54 Votes: Aligning Africa’s Position on Global Compute Access 

From the multilateral perspective, the challenge is translating internal alignment into external bargaining power. Mehdi Snene, Senior Advisor to the UN Secretary-General’s Special Envoy for Digital and Emerging Technologies, reminded the consultation that 80% of African nations still lack a formal national AI strategy, making continental coordination through the AU essential.

Snene emphasized that Africa’s 54 UN member states represent over a quarter of the voting block in global governance bodies, a position of immense leverage if unified around clear, non-negotiable infrastructure demands.

“Africa represents 54 member states, more than a quarter of the United Nations membership. That gives us a very strong voice provided it comes together. A coherent voice carries a small number of clear asks: equitable access to compute, access to data and talent, predictable financing, and meaningful participation in international standard-setting bodies.”

– Mehdi Snene, Senior Advisor, UN SG’s Office for Digital and Emerging Technologies

It’s the same argument Wamola made about infrastructure, translated to the multilateral stage: a unified continental position, not 55 separate ones, is what gives Africa something to negotiate with. Snene urged African delegations to prepare a unified position ahead of the second UN Global Dialogue on AI Governance in New York, positioning the continent as an active architect of global frameworks rather than a recipient of external rules.

Real-World Deployment: Moving from Model Laws to Operational Tools 

From the research and technical side, panelists warned against drafting theoretical frameworks that fail upon contact with real-world operating environments.

Dr. Girmaw Abebe Tadesse, Principal Scientist at Microsoft AI for Good Lab and member of the UN Scientific Panel on AI, argued that African AI governance should focus on domain-specific decision-making, such as agriculture, health, and local finance, rather than chasing generic general-purpose models. He stressed the importance of context-specific testing:

“A solution built in Lagos assuming that context, when tested somewhere in Rwanda or rural Nigeria, may fail because we don’t often test those solutions in their actual operating environment. Beyond building technology, we must evaluate AI systems under the actual conditions where they will be deployed.”

– Dr. Girmaw Abebe Tadesse, Principal Scientist, Microsoft AI for Good Lab

This concern was echoed during the open consultation by public sector practitioners who urged the AU to provide regulators with practical toolkits rather than abstract declarations. Robert Ifeonu, Digital Development Specialist at the Central Bank of Nigeria, highlighted the execution gap facing public institutions:

“My fear is for these documents to become well-articulated strategies that cannot be implemented in real environments. Institutions are struggling with simple use-case selection and managing trade-offs. We need to give governments actual tools they can use. What does an AI governance framework mean for a central bank? What does that mean for a tax authority?”

– Robert Ifeonu, Digital Development Specialist, Central Bank of Nigeria

Between them, Tadesse and Mfongo made the case that Wamola’s continental baseline only works if it comes with something to implement, not just something to sign.

The Next Frontier: What Africa Chooses to Govern Together 

As the African Union prepares its draft guidelines and model law for ministerial review, the consultation boiled down to a single question: which parts of AI governance are too important to leave fragmented across national borders, and which are best handled locally?

Wamola’s answer, delivered near the panel’s close, doubled as a working principle for the consultation as a whole: harmonization doesn’t mean uniformity. It means every country moving toward the same destination without waiting for the slowest, or the fastest, to set the pace.

“We’re not talking about imposing identical legislation. What we’re talking about is continental baseline, interoperable flows and standards that will allow every country, depending on where they are on their maturity level of frontier technologies adaptation, to progressively come towards harmonizing all that we’re talking about here in terms of AI governance. Everyone at their own speed, but remaining aligned to the same continental direction on what we want to achieve.”

– Angela Wamola, Head of Africa, GSMA

That, more than any single clause likely to appear in the draft guidelines, is the test the AU’s model law will have to pass: not whether it regulates AI, but whether it lets 55 countries move together without needing to move identically.