Cybersecurity Education Initiative (CYSED)

For the Global South, the first Global Dialogue on AI Governance will only be a success if it actively shifts the international focus from "Layer 2" policy governance to "Layer 1" foundational governance. Currently, developing nations are trapped in a reactionary cycle, drafting regulations for imported, closed-source models they did not build and cannot fully audit. True governance is impossible without the technical capacity to develop the technology. Therefore, a successful Dialogue must deliver three concrete outcomes: 1. Recognition of the Compute Gap as a Governance Crisis: The Dialogue must formally acknowledge that the lack of high-performance computing (HPC) in the Global South is not merely an economic disadvantage, but a critical governance failure. Regulating an imported black box is an illusion of governance; true governance requires local ownership. 2. Actionable Infrastructure Partnerships: Success means moving beyond high-level ethical declarations to establish tangible, multistakeholder funding mechanisms. The Dialogue should yield a commitment to a "Global South Compute Partnership" aimed at financing localized HPC clusters and supporting open-source education for grassroots innovators, such as those championed by CYSED and Africa Cyberfest. 3. Endorsement of Indigenous Engineering: The inaugural report of the Independent International Scientific Panel on AI must explicitly outline strategies for scaling "indigenous engineering." It must prioritize transitioning African youth from passive consumers of foreign technology into active architects of culturally relevant, secure AI systems. Ultimately, the July Dialogue will be a success if it recognizes that in AI, code is law. The ultimate outcome must be a global commitment to democratizing the infrastructure required to build AI, ensuring that the capacity to govern AI is as globally distributed as its impact.

• Open-source software, open data and open AI models
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• AI capacity-building

The selected priorities represent the critical path for the Global South to achieve digital sovereignty and transition from passive AI consumers to active creators. AI Capacity-Building: This is our most urgent priority. Before developing nations can effectively govern AI, we must be able to build it. Addressing the global "compute gap" by funding local high-performance computing (HPC) and indigenous engineering education is the prerequisite for all other governance efforts. Open-Source Software, Data, and Models: Open-source AI is the ultimate equalizer for developing digital economies. It democratizes access to foundational technology, allowing grassroots innovators to build solutions without starting from scratch. However, these open models are only effective if local developers have the infrastructure to run and fine-tune them. Social, Cultural, and Linguistic Implications: Imported, closed-source models inherently carry the biases of their creators and frequently fail to represent African languages, cultural nuances, and socio-economic realities. To ensure AI serves our communities ethically, we must develop localized models that reflect our unique context rather than relying on imported assumptions. Transparency, Accountability, and Human Oversight: True accountability is impossible when relying on foreign "black box" systems. We cannot effectively audit or oversee algorithms we did not build. Transparency requires "Layer 1" foundational governance-meaning local engineers must be involved in the coding, training, and deployment phases to ensure systemic integrity. Together, these priorities form the foundation of our work at CYSED. They underscore that equitable global AI governance must be rooted in democratized capacity and culturally representative engineering.

Yes, a critical cross-cutting issue absent from the listed themes is the Physical Infrastructure and Compute Divide, and the resulting Governance Paradox it creates for the Global South. Current governance frameworks treat AI primarily as a software, data, and policy issue, largely ignoring the physical hardware, specifically, equitable access to high-performance computing (HPC), GPUs, and energy infrastructure. For developing nations, this compute bottleneck is the ultimate barrier. We cannot operationalize "open-source models" or achieve true "capacity-building" without the physical infrastructure to train and host these systems locally. This physical divide creates an emerging crisis: The Illusion of Second-Level Governance. AI governance is effectively splitting into two tiers: Layer 1 (Foundational Governance): Dictated by the creators who own the compute, baking governance directly into the model weights, algorithms, and training data. Layer 2 (Policy Governance): Relegated to consumers in the Global South, who spend years drafting regulations for imported, closed-source models they did not build. Regulating an imported "black box" is not true governance; it is merely playing defense against a finished product. If a nation only controls the "use" phase of an AI model, it has zero governance over its core architecture. Therefore, the democratization of physical AI infrastructure must be recognized as a distinct, cross-cutting governance issue. Without equitable access to compute power, the Global South will remain permanently trapped in Layer 2, unable to practice "indigenous engineering." True digital sovereignty can only be achieved when the hardware capacity to build AI is as globally distributed as the policies designed to regulate it.

In Nigeria and the broader West African region, the gap between foundational AI development and policy governance presents both a severe challenge and an unprecedented opportunity. The Significant Challenges: The most pressing challenge is our tech ecosystem's heavy reliance on imported, closed-source AI models. Because we face a severe high-performance computing (HPC) deficit, our region is forced into a state of digital dependency. This creates profound socio-cultural and technical vulnerabilities. Foreign models frequently fail to comprehend local languages, such as Yoruba, Hausa, or Igbo, and lack context for our unique socio-economic realities, leading to biased solutions. Furthermore, as I have demonstrated regarding AI-enhanced social engineering, the weaponization of AI poses an asymmetric cybersecurity threat to our region. We cannot effectively combat these threats, nor ensure transparency and accountability, using "black box" algorithms we do not control. The Opportunities: Conversely, our greatest opportunity lies in our demographic dividend. Africa possesses the youngest population globally, a massive, untapped reservoir of technical talent. The proliferation of open-source software and open AI models provides the foundational building blocks for innovation. If the international community partners with grassroots initiatives like CYSED and Africa Cyberfest to bridge the physical compute gap, we can catalyze a regional movement of "indigenous engineering." By shifting our youth from consumers to creators, we have the opportunity to build culturally representative, secure AI systems tailored to local needs. Empowering African developers to operate at "Layer 1" (foundational governance) will not only secure our digital economies but also ensure the global AI ecosystem is truly inclusive, accountable, and equitable.

The Global Dialogue on AI Governance has the unique potential to redefine international cooperation, shifting it from the mere exportation of regulatory frameworks to the equitable distribution of foundational AI capacity. To truly advance global cooperation, the Dialogue must play three critical roles: 1. Reframing Governance as a Capacity Issue: The Dialogue can disrupt the current illusion of second-level policy governance. By officially acknowledging that true governance requires the technical ability to build and audit AI (Layer 1), the Dialogue can redirect international cooperation toward bridging the global compute and capacity gap, rather than focusing solely on risk compliance. 2. Brokering Actionable Infrastructure Partnerships: Instead of functioning solely as a policy forum, the Dialogue must serve as a catalyst for tangible investment. It can broker multistakeholder partnerships that connect Global North resources with Global South talent. By facilitating "Global South Compute Partnerships," the Dialogue can secure funding for regional high-performance computing (HPC) clusters and support grassroots educational hubs like CYSED and Africa Cyberfest. 3. Democratizing Open-Source Ecosystems: International cooperation must move beyond shared ethics to shared assets. The Dialogue can drive global agreements that protect and promote open-source AI models, ensuring they remain accessible to developers in developing nations as the ultimate equalizer for local innovation. Ultimately, the AI Dialogue's most vital role is ensuring the Global South is not relegated to being passive rule-takers and technology consumers. By championing "indigenous engineering" and facilitating infrastructural cooperation, the Dialogue can ensure all nations have the sovereign capacity to architect safe, contextually relevant AI.

To maximize impact, the AI Dialogue must bridge the gap between high-level international policy and grassroots technical execution. It should actively build upon and connect with: 1. Regional and Grassroots Initiatives: The Dialogue must engage with mechanisms like the African Union's Continental AI Strategy and, crucially, grassroots platforms such as the Cybersecurity Education Initiative (CYSED). These initiatives are already doing the groundwork of training African youth in indigenous engineering and cybersecurity. 2. UN and Global Frameworks: It should build upon UNESCO's Recommendation on the Ethics of AI and the ITU's AI for Good platform. While these have established strong "Layer 2" policy and ethical baselines, they currently lack the physical infrastructure mechanisms required for true implementation in developing regions. 3. Open-Source Tech Alliances: Global networks championing open-source models must be integrated, as they provide the foundational building blocks for local developers. The Added Value of the AI Dialogue: The Dialogue's unique added value lies in acting as the ultimate integration layer. Currently, global mechanisms are siloed: policymakers discuss ethics at the UN, while engineers in the Global South struggle for compute access. The Dialogue can synthesize these efforts by tying policy directly to infrastructure. Instead of reinventing ethical guidelines, the Dialogue can operationalize them by brokering tangible "Global South Compute Partnerships." By connecting the high-level backing of the UN with the grassroots execution capabilities of entities like CYSED, the Dialogue transforms abstract AI governance into tangible AI capacity—providing the physical hardware required for developing nations to participate in Layer 1 foundational governance.

To move beyond theoretical policy, the AI Dialogue must restructure how stakeholders engage, ensuring contributions translate into physical capacity-building in the Global South. Stakeholder Contributions: Private Sector & Tech Giants: Must move beyond exporting closed-source models or merely offering "API access." They should contribute tangible infrastructure such as high-performance computing (HPC) subsidies—and open-source models to empower local developers. International Funders & Member States: Should pivot from funding mere digital awareness campaigns to financing "Global South Compute Partnerships," directly subsidizing the hardware required for indigenous AI development. Grassroots Organizations (e.g., CYSED): Must serve as the implementation layer, identifying local talent and driving the "indigenous engineering" education required to build culturally relevant, secure AI. Recommendations for Format and Structure: The July AI Dialogue must disrupt the traditional UN format of plenary sessions focused purely on "Layer 2" policy. We recommend the following structural changes: 1. Dedicated "Compute and Capacity" Track: Elevate physical infrastructure from a footnote to a primary thematic pillar, exploring actionable mechanisms to democratize HPC in developing nations. 2. "Layer 1" Technical Inclusion: Ensure panels are not dominated solely by diplomats and lawyers. The dialogue must center technical architects and grassroots engineers from the Global South who actually understand model development. If code is law, engineers must be at the governance table. 3. Action-Oriented Partnership Hubs: Transition the Dialogue from a "talking shop" to a matchmaking platform. Incorporate structured sessions that directly connect global tech entities and donor agencies with grassroots engineering hubs to finalize infrastructure funding commitments. By adopting this structure, the Dialogue ensures stakeholders are actively building the capacity for all nations to participate in true AI governance.

The most glaringly underrepresented voices in global AI governance are the grassroots technical builders—the developers, engineers, and youth of the Global South. Current global discussions are overwhelmingly dominated by Global North policymakers, tech conglomerates, and legal experts focusing on "Layer 2" regulation. Meanwhile, the perspectives of those facing the severe "compute gap" in developing nations are sidelined. Furthermore, non-Western linguistic and cultural communities are excluded; because foundational models are primarily trained on Western datasets, they remain culturally blind to regions like West Africa. This exclusion creates a paradox: the communities most vulnerable to AI biases and AI-enhanced cyber threats have the least agency in shaping the technology's foundational architecture ("Layer 1" governance). To meaningfully include these voices, the AI Dialogue must democratize the table: 1. Elevate Grassroots Hubs: Organizations like CYSED and Africa Cyberfest must be recognized as primary implementation partners. We are already on the ground, transitioning African youth from technology consumers to creators through "indigenous engineering." 2. Fund Physical Inclusion: We cannot invite Global South engineers to the AI development table without providing the physical tools. International stakeholders must establish "Global South Compute Partnerships" to fund localized high-performance computing (HPC) clusters and open-source education. 3. Center Technical Architects: Governance forums must explicitly invite engineers from developing nations, not just diplomats. If code is law, the people writing the code must reflect the global majority. True inclusion means empowering the Global South with the sovereign capacity to build AI, not just the mandate to comply with it.

To break the cycle of traditional UN plenary sessions which often reinforce the "Layer 2" policy charade the AI Dialogue must adopt formats that reflect the technical and infrastructural realities of "Layer 1" AI development. We recommend three innovative engagement formats to foster dynamic, meaningful outcomes: 1. Live "Layer 1" Technical Showcases: Theoretical discussions on AI risk are insufficient. The Dialogue should feature live technical demonstrations led by Global South engineers. Showcasing real-world vulnerabilities—such as the AI-enhanced social engineering threats we have demonstrated at Africa Cyberfest and GITEX—or conducting live red-teaming of open-source models will ground abstract policy discussions in immediate technical realities. 2. Infrastructure Matchmaking Hubs: To move beyond high-level declarations, the Dialogue should implement structured, action-oriented "deal rooms." These hubs would facilitate direct matchmaking between tech conglomerates, donor states, and grassroots implementation partners like CYSED. The explicit goal would be to negotiate and secure "Global South Compute Partnerships," translating governance commitments directly into funded, localized high-performance computing (HPC) clusters. 3. "Code is Law" Design Sprints: Instead of isolated policy panels, host collaborative design sprints where diplomats, legal experts, and Global South developers sit at the same table. The objective would be to instantly stress-test proposed regulatory frameworks against actual open AI models. This ensures that policies drafted in Geneva are technically feasible and culturally applicable in regions like West Africa. By shifting from passive listening to active technical collaboration and infrastructure matchmaking, the Dialogue will yield tangible capacity-building outcomes.

To effectively address the challenges of AI governance in the Global South, we must champion approaches that solve the "Layer 1" foundational capacity gap. Existing frameworks often stall because they lack implementation mechanisms for developing nations. The most effective concrete solutions combine localized education with infrastructural access. 1. Grassroots Capacity Platforms (CYSED & Africa Cyberfest): Our work at the Cybersecurity Education Initiative (CYSED) and the Africa Cyberfest platform serves as a prime example of actionable governance. Rather than waiting for top-down regulation, we actively build foundational capacity. By training West African youth in AI-enhanced cybersecurity and "indigenous engineering," we offer a concrete solution to digital dependency. We ensure local talent can audit, secure, and build contextually relevant models, which is the absolute prerequisite for true accountability and human oversight. 2. Open-Source AI Ecosystems: Approaches that champion open-source models and decentralized datasets are vital governance practices. They prevent the monopolization of AI by a few global tech giants and act as the ultimate equalizer for grassroots innovators. However, open-source is only half the solution; it must be coupled with the hardware to run it. 3. Subsidized Sovereign Compute Policies: A concrete policy approach the UN must champion is treating high-performance computing (HPC) as a global public good. Establishing "Global South Compute Partnerships"-where international bodies and the private sector co-fund regional compute clusters-is a tangible governance solution. This practice directly addresses the hardware bottleneck, allowing developing nations to train culturally representative models locally rather than relying on imported, biased systems. Ultimately, the most effective AI governance approach is democratizing the means of production. Shifting the Global South from passive compliance to active AI creation is the ultimate regulatory safeguard.