AfriTechElles

A successful first Global Dialogue on AI Governance would, in my view, be one that moves beyond declarations and produces actionable outcomes that reflect the realities of regions like Africa. For the Dialogue to be truly meaningful, it must acknowledge the critical role that Digital Public Infrastructure plays as the foundation for AI innovation — and recognize that many African nations have already made significant investments in DPI systems. A successful outcome would include concrete frameworks that guide how this infrastructure data can be responsibly leveraged for AI development, with clear safeguards around privacy, consent, and equity. Regional readiness must also be central to the conversation. Success would mean the Dialogue produces recommendations that empower regional bodies like the African Union to lead context-specific AI governance efforts, rather than simply adopting frameworks designed elsewhere. Africa's governance challenges and opportunities are distinct, and the outcomes should reflect that. Finally, a successful Dialogue would result in a commitment to inclusive participation — ensuring that civil society organizations, women-led tech communities, and underrepresented stakeholders from the Global South are not just consulted but genuinely heard. AI governance that does not center those most impacted by these systems will ultimately fall short. In short, success looks like a Dialogue that produces governance frameworks that are regionally adaptable, grounded in existing digital infrastructure realities, and inclusive enough to leave no one behind.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• AI capacity-building
• Transparency, accountability, and human oversight

As an organization working at the intersection of technology, gender, and digital development in Africa, these four thematic areas directly reflect our priorities and the realities we navigate daily. AI capacity-building is urgent for Africa. Many countries on the continent have invested significantly in Digital Public Infrastructure, yet the human capital and institutional capacity to harness that infrastructure for responsible AI innovation remains limited. Without deliberate capacity-building efforts, Africa risks being a data source rather than an AI beneficiary. The social, economic, ethical, cultural, linguistic and technical implications of AI are particularly acute in African contexts. AI systems trained on non-representative data can deepen existing inequalities, especially for women and marginalized communities. Governance frameworks must account for Africa's diverse linguistic landscape and socioeconomic realities. Interoperability of governance approaches is essential for regional readiness. Africa is not a monolith - it comprises 54 countries with varying levels of DPI maturity and regulatory capacity. For AI governance to work at scale, frameworks must be interoperable across borders, enabling regional bodies like the African Union to coordinate effectively without forcing a one-size-fits-all approach. Finally, transparency, accountability, and human oversight are non-negotiable when DPI data - which includes sensitive identity, health, and financial information - is leveraged for AI development. Communities must be able to trust the systems built on their data, and meaningful oversight mechanisms must be in place to ensure accountability. Together, these priorities reflect our commitment to an AI governance framework that is inclusive, regionally grounded, and built on trust.

Yes, and I think it's an important gap to name clearly. One issue that doesn't get enough attention is how existing Digital Public Infrastructure data intersects with AI development, especially in Africa. Many African countries have made real investments in DPI - national identity systems, payment platforms, health registries, agricultural databases - yet there is no clear international guidance on how this data can be responsibly used for AI training and innovation. That absence creates both serious risks and missed opportunities that the current themes don't fully capture. Related to this is the question of gendered data gaps. Women across Africa are still underrepresented in many digital systems, from identity registration to financial services. When AI is built on data that already reflects those gaps, it tends to reproduce and deepen existing inequalities. This goes beyond general ethics discussions and deserves specific, dedicated attention. There is also the issue of who actually owns and controls AI infrastructure. Compute power, foundational models, and cloud systems are concentrated in the hands of very few actors in very few countries. African nations that generate valuable DPI data often have little say in how that data is ultimately used. Any serious governance framework needs to confront this structural imbalance. Finally, the vast majority of AI tools are built for a small number of high-resource languages, leaving most of Africa's population unable to interact with or benefit from these systems in their own languages. Multilingual inclusion is not a niche concern - it is a fundamental equity issue. These are cross-cutting challenges that touch every theme on the list and deserve a dedicated place in the Dialogue.

Across Africa, the governance gaps in AI are not abstract policy problems — they are lived realities that affect millions of people who do not even know it yet. A few months ago, I was one of them. I had no idea what Digital Public Infrastructure was, or why AI governance mattered to communities like mine. Today, having engaged deeply with these issues, I feel a profound responsibility to bring this knowledge to every woman, every young person, every community that is still where I was — unaware that decisions being made in global forums will shape their lives in very real ways. This is perhaps the most significant challenge across Africa: the awareness gap. AI governance discussions are happening at high levels while the majority of Africans — especially women — remain disconnected from these conversations, unaware of how their data, their identities, and their futures are being shaped by systems they had no voice in designing. At the same time, the opportunity is enormous. Africa has a young, growing population, expanding DPI systems, and a wealth of data that could power AI innovation tailored to African realities. But without proper governance frameworks, that potential will be captured by others rather than benefit African communities themselves. The capacity-building gap is real. The transparency gap is real. The interoperability gap between African nations is real. But so is the hunger for knowledge and participation once people are given access to it. The work is not just policy — it is awareness, education, and inclusion. And that work starts now.

Honestly, the AI Dialogue comes at a moment where the world really needs it. For a long time, international cooperation on AI governance has mostly meant that powerful countries set the rules and everyone else tries to keep up. The Dialogue has a real chance to break that pattern — but only if it takes inclusion seriously, not just as a principle but as a practice. What I think the Dialogue can uniquely do is create a space where countries learn from each other rather than just from the top down. Africa, for example, has built DPI systems that many developed nations are only beginning to explore. That experience is valuable. South-South cooperation and peer learning between nations at similar stages of development is something the Dialogue is perfectly placed to encourage and formalize. The Dialogue can also serve as a bridge between what regional bodies like the African Union are already building and what global standards are emerging. Right now those conversations are happening in parallel. They need to be in dialogue with each other — which is exactly what this platform is for. But what excites me most is the potential to make AI governance something that ordinary people actually understand and engage with. A few months ago I had no idea these conversations were happening. Now I do, and I feel the urgency to make sure others do too. The Dialogue can help create that awareness at scale — building a shared language around AI that reaches beyond experts and diplomats into communities that are most affected. That, to me, is what meaningful international cooperation looks like.

There are already some really important efforts underway that the AI Dialogue should not ignore or duplicate — it should connect with them and build on what they have already learned. The African Union's Continental AI Strategy is an obvious starting point. The AU has been doing the hard work of thinking through what AI governance means in an African context, and that work deserves a seat at the global table — not just as a reference document but as a living input into the Dialogue's outcomes. The ITU's AI for Good platform, the UNDP's work on digital transformation, and UNESCO's Recommendation on the Ethics of AI are all initiatives that have laid important groundwork. The Dialogue should actively connect with these rather than starting from scratch. On the DPI side, initiatives like the GovStack framework and the work of the Co-Develop Fund have been helping countries build and govern digital public infrastructure in ways that are open and interoperable. Since DPI is increasingly the foundation on which AI systems will be built, especially in the Global South, the Dialogue cannot afford to treat AI governance in isolation from these efforts. Civil society networks — including women-led tech organizations working across Africa — have also been quietly doing the awareness and advocacy work that formal institutions often cannot reach. The Dialogue should see these networks as partners, not just stakeholders to consult. The added value the Dialogue brings is something none of these initiatives have on their own — a truly global, UN-backed platform where all of these threads can come together, find common ground, and produce recommendations that carry real political weight. That convening power is rare. It should be used wisely.

The Dialogue will only be as strong as the diversity of voices that shape it. Governments bring policy authority, but they cannot do this alone. Civil society organizations bring ground-level insight that no government report can fully capture. The private sector brings technical knowledge and resources. Academia brings evidence and long-term thinking. And the technical community understands the systems well enough to know where the real risks lie. What I would recommend is that the Dialogue move away from the traditional format where governments speak and everyone else observes. Each stakeholder group should have structured, meaningful opportunities to contribute — not just during open floors but in the actual drafting of outcomes. Civil society and women-led organizations in particular should be resourced to participate, not just invited. The format should also allow for written inputs from communities that cannot travel to Geneva or New York. A global dialogue should not be limited to those who can afford to be in the room.

Honestly, the list is long — but a few stand out. Women, particularly from the Global South, are almost invisible in these conversations despite being among those most affected by AI systems. Rural communities, persons with disabilities, indigenous populations, and speakers of low-resource languages are rarely at the table. Young people are another — they will live longest with the consequences of today's governance decisions, yet they are consistently sidelined. Including these voices requires more than translation services and livestreams. It requires active outreach, local consultations, and building trust with communities that have historically been spoken about rather than spoken with. Organizations like ours exist precisely to bridge that gap — and the Dialogue should lean on us to do so.

The standard panel-and-plenary format will not be enough to generate the kind of dynamic, honest exchange this moment calls for. I would love to see the Dialogue experiment with community listening sessions held in different regions before the Geneva meeting — so that what happens in the room actually reflects what people on the ground are saying. These could be facilitated by local civil society organizations and the findings fed directly into the agenda. Deliberative dialogue formats — where participants with different perspectives work through a real problem together rather than presenting prepared positions — can also generate much richer outcomes than traditional debates. Digital participation tools that allow people to contribute in multiple languages, asynchronously, would open the Dialogue to voices that cannot engage in real time. And storytelling formats — where affected communities share their experiences directly — can make abstract governance discussions feel urgent and human in a way that policy papers rarely do. The Dialogue should also consider youth-specific engagement tracks. Not a side event, but a genuinely integrated space where young people's perspectives feed into the main outcomes. The goal is a Dialogue that people feel belongs to them — not one they are watching from the outside.

There are some really encouraging examples out there that show what effective AI governance can look like when it is grounded in real context. India's approach to Digital Public Infrastructure through the India Stack - Aadhaar, UPI, and the Data Empowerment and Protection Architecture - offers genuinely useful lessons about how to build open, interoperable systems that can eventually support responsible AI development at scale. It is not perfect, and the privacy debates around it are important, but the model of government-backed open infrastructure that private and public actors can build on is one Africa is already learning from. Rwanda's National AI Policy is one of the clearest examples on the African continent of a government taking AI governance seriously in a way that reflects local priorities rather than simply copying external frameworks. It is practical, it is grounded, and it is worth building on. The EU AI Act, whatever its limitations, has demonstrated that it is possible to create binding regulatory frameworks for AI that are risk-based and sector-specific. The conversation it has sparked globally - including in Africa - has been valuable even for those who will not adopt it directly. On the civil society side, initiatives like the Feminist AI network and various digital rights organizations across Africa have been developing community-centered approaches to AI accountability that formal governance bodies rarely achieve on their own. And at the infrastructure level, the GovStack initiative is showing how DPI can be built in a way that is open, reusable, and governance-ready from the start - which is exactly the kind of foundation responsible AI development needs. These are not perfect models. But they are honest starting points the Dialogue should engage with seriously.