In a region with differing levels of AI policy maturity and highly varied levels of inequality, mitigating AI’s risks – which touch everything from freedom of speech to political trust, surveillance, child safety and election integrity – requires more than simply having policies in place. To truly move the needle across all high-risk areas, the best option often becomes the most difficult: regional coordination.
While one country is well-equipped to manage and fund AI’s impacts on national defence, another will be better resourced to tackle water consumption and data centre expansion. Together, however, both can learn to apply best practices, cohesively regulating tools whose effects rarely stay within country borders anyway. As Patricia Ainembabazi put it during the Africa Internet Governance Forum’s session, ‘Governing the Code: Comparative AI Lessons from East & Southern Africa’, hosted on 3 November 2026 in Accra, Ghana: “When you want to learn to drive, you learn from someone who drives.”
Moderated by Peace Amuge (Association for Progressive Communications), the panel sought to compare and contrast differing national approaches to mitigating the cumulative risks of ‘frontier technology’, and was joined by Mpho Moyo (Research ICT Africa), Edrine Wanyama (CIPESA), Muthuri Kathure (BBC Media Action) and Patricia Ainembabazi (CIPESA).

With instruments such as the African Union Data Policy Framework, Continental AI Strategy, and the Digital Transformation Strategy for Africa available, most panellists agreed the real challenge lay in implementation and harmonisation, not policy drafting and standard-setting. But global pressures often rush nation-states to regulate before they’re ready to put policies into practice.
One audience member asked, “We strive to bridge the digital divide, but we need electricity for computers, data and AI. While we seek to govern the technology, we still have issues of electricity, connectivity and access for everyone, whether in Tanzania, Kenya, Ghana or elsewhere in Africa. Are we just rushing to govern?”
Citing three of Research ICT Africa’s analyses of national AI policies through a Just AI lens, Moyo noted that, when it came to AI policy development, each country had achieved some success in aligning governance commitments with local priorities. The problem, however, was finding the institutional mechanisms and peoplepower to enforce policies, not to mention create the conditions for local buy-in and benefit distribution.
“We found that Egypt focused on investment, infrastructure, and building skills, all important foundations for AI development. But it still needs stronger human rights safeguards, public participation and independent accountability. In Zambia, their strategy calls for an AI council. But will marginalised communities have any influence in those arrangements? When we say “we consulted,” does that translate into meaningful participation and benefit? Likewise, Mauritius has an interesting approach centred on investment, productivity, and competitiveness, but it does not quite ask who will be excluded, whose livelihoods are affected, or who bears the environmental costs of developing these systems.”
Without access to basic services such as electricity or water, one panellist asked what good AI could do for local communities. Even with the best policies in place, without supporting conditions such as education, connectivity, and access to basic redress mechanisms (for when things inevitably go wrong), the group pointed out that AI governance strategies and policies can often seem obsolete.
Wanyama said, “Consider a marginalised community where nobody speaks English, or can read and write, and you introduce AI concepts to them. How do consultation and effective participation work in practice? Language is a barrier, and the language used for complex AI matters is difficult for non-technical people. Likewise, in many African societies, people are more worried about what they will eat next.”
Frequent food shortages, the converging crises of the cost of living, and other immediate challenges tend to eclipse concerns about AI’s impact on abstract issues like data privacy or surveillance. But as Wanyama argued, “I foresee that AI will become very critical to our discussions, and eventually, even the smallest person in the village will know about it. This is because it affects the environment: it uses water to cool servers that generate content, and those servers emit carbon dioxide. One day, our crops and our food will be affected. These impacts will create problems for everyone and have lasting impacts beyond just what I am eating today or tomorrow.”
The challenge then becomes creating awareness about AI’s long-term impacts on digital and human rights. How then can AI governance be shaped and redesigned with local communities’ immediate challenges in mind? Panellists argued that implementation plans must consider how technologies will land with real people living at the intersection of multiple overlapping inequalities. Until then, AI deployment in Africa will remain inequitable.
Moyo said: “We need to ask whether women, rural communities and persons with disabilities can shape AI priorities. Do these systems recognise our languages and knowledge systems? Can we make real decisions about what is done with our data? We should not simply say we have a strategy or a policy, because that is only the beginning. What matters is how it translates into reality from an African perspective.”
Wanyama agreed, saying: “One of the most striking discussions, cutting across every group (technical community, civil society, government), was the inclusion of marginalised groups… people think of persons with disabilities, but we are also talking about your grandmother, grandfather and cousins who live in the village. We are talking about families where technology is more accessible to the man than to the mother and children… That comes down to inclusivity, and with it, participation.”
On this note, Amuge spoke to public participation and multistakeholder engagement, particularly with civil society: “Capacity is very important, and so is collaboration. People are often unaware of these processes and how to take part. But when civil society organisations and partners come together to make input, it is stronger.”
Ultimately, the session emphasised that AI governance requires collaboration and capacity building to truly reflect the diversity of African needs. While the policy instruments already exist, their value must be measured by institutions’ ability to enforce them, the foundational conditions shaping their usage and adoption, and whether the most marginalised can have a say in shaping the governance priorities meant to serve them. As Ainembabazi said: “Good systems are built when you and I, the owners of the data, share our experiences from our countries and ask the right questions.”
Regional coordination around policy implementation offers a way forward, allowing countries to pool resources, share lessons and, more importantly, share the burden of holding global technology companies to account – but ultimately, that coordination only matters if the policy can translate into meaningful benefit, and justice, for all those it claims to serve.