RTivara Advisory

Success cannot be measured by the ambition of language alone. From a legal and accountability standpoint, three outcomes matter. First, a co-chair summary that maps genuine convergence, not the lowest common denominator of the most reluctant participants. The Dialogue is deliberately designed as a non-negotiating forum, concluding with a co-chair summary rather than a negotiated text. That design flexibility is an asset, but only if the summary reflects substantive African priorities across all seven thematic areas, rather than deferring to the positions of the most technologically dominant states. Second, visible and structural Global South participation in the process, not just its outcomes. For African legal and regulatory communities, success means African voices shaping the co-chair facilitation process itself, African experts represented on the Independent International Scientific Panel, and African regulatory frameworks, including the AU Continental AI Strategy and AU Data Policy Framework, explicitly acknowledged in the summary as relevant governance instruments. Third, a credible bridge to 2027. The first Dialogue is a foundation, and not a conclusion. Success means leaving Geneva with an agreed set of priority issues for the New York session, a commitment to better-resourced Global South participation in preparatory consultations, and a process for the Scientific Panel's annual report to feed directly into governance decisions, not simply inform general discussions. From an AI assurance perspective, a successful Dialogue also establishes a baseline: a shared understanding of what accountability for AI systems deployed across borders actually requires in practice. Without that baseline, the seven thematic areas remain disconnected aspirations rather than components of a coherent governance architecture.

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

All seven areas have merit. Four are urgent for Africa in the immediate term, and their urgency is structural rather than merely political. AI Capacity-Building is the foundational priority. Africa accounts for approximately 2% of global data centres. Without sovereign computational infrastructure, African states cannot build, audit, or regulate AI systems affecting their populations. Capacity-building commitments must be time-bound, measurable, and African-led, not externally designed assistance delivered on donor terms. This is a precondition for a meaningful participation in every other thematic area. The Societal, Cultural and Linguistic Implications thematic area is the most distinctively African in character. Africa's 2,000-plus languages are systematically excluded from the foundational models being deployed on the continent. This is not a cultural observation; it is a legal and accountability failure. AI systems making consequential decisions in healthcare, agriculture, and justice cannot be accountable if the people they affect cannot access, understand, or challenge them in their own languages. Transparency, Accountability and Human Oversight is the area where my professional expertise is most directly engaged. Cross-border accountability gaps are currently unaddressed by any binding framework. Foreign-built AI systems deployed in African contexts operate in a legal vacuum. The Dialogue must produce a shared understanding, at minimum, of what meaningful human oversight requires when the developer, deployer, and affected population are in different jurisdictions. Interoperability of Governance Approaches is urgent because without it, African governance frameworks are de facto overridden by frameworks built for other contexts. The AU Continental AI Strategy, the Kigali Declaration on AI, and the AfCFTA Digital Trade Protocols represent a coherent African governance architecture. The Dialogue must ensure that global interoperability standards are compatible with, not dismissive of, this body of work.

The consultation explicitly invites identification of gaps. From a legal and AI assurance perspective, four cross-cutting issues fall outside the current thematic architecture and are of particular relevance to African states. AI in Armed Conflict and Autonomous Weapons. This is the most glaring structural omission. The automated weapons market in Africa has grown to approximately $1 billion in imports. The use of AI-enabled drones in the Sahel and Horn of Africa is already documented. None of the seven thematic areas provides a space to address lethal autonomous weapons systems (LAWS) or the application of international humanitarian law to AI-enabled targeting. 120 UN Member States support a new LAWS treaty. The Dialogue cannot credibly address safe AI while leaving its most dangerous application unaddressed. AI, Labour Markets and Economic Displacement. The World Economic Forum's Global Risks Report 2026 identifies adverse AI outcomes as the risk with the largest rise in ranking over the past decade. For Africa, where informal labour markets are dominant and worker protections are limited, AI-driven displacement has no existing governance framework. The seven thematic areas contain no dedicated just-transition or labour rights mechanism. AI, Misinformation and Information Integrity. Misinformation and disinformation rank second on the two-year global risk outlook. For African democracies facing election cycles, AI-enabled disinformation is a rule-of-law issue as much as a technical one. The relationship between information integrity as a governance objective and freedom of expression as a legal constraint is structurally absent from the seven thematic areas. AI, Gender and Technology-Facilitated Violence. The AU Peace and Security Council's March 2026 session addressed AI's implications for women, peace and security, including technology-facilitated violence. Gender is neither a standalone theme nor an explicitly mandated cross-cutting lens in the current Dialogue framework. This gap must be corrected in the co-chair summary and addressed as a structural priority in 2027.

Africa's engagement with artificial intelligence presents both structural vulnerabilities and strategic opportunities across the four key domains I earlier highlighted. First, in AI capacity-building, Africa remains underrepresented in global infrastructure and talent, hosting only 2% of data centres and 3% of the AI workforce. This imbalance heightens exposure to technological dependency, exploitative data practices, and limited oversight capacity, particularly as independent regulatory institutions remain largely absent. However, rapid market growth, projected at 17.5% in 2025, alongside major investments and emerging sovereign infrastructure projects, signals a critical window for establishing governance systems aligned with the African Union (AU) Continental AI Strategy. Second, societal, cultural, and linguistic considerations highlight systemic underrepresentation of African languages in AI systems, driven by weak data ecosystems and foreign-dominated datasets. This risks cultural erasure and reinforces external dependency. Yet, national initiatives in countries such as Nigeria, Ethiopia, and Senegal, alongside community-driven models like Masakhane, demonstrate the potential for locally grounded, multilingual AI systems that treat linguistic data as sovereign infrastructure. Third, transparency, accountability, and human oversight remain fragmented. Divergent national approaches and reliance on foreign-built systems have created regulatory gaps and limited cross-border accountability, particularly affecting vulnerable populations. Nonetheless, growing judicial engagement, strategic litigation, and AU-led governance efforts provide a foundation for developing a coherent continental accountability architecture, contingent on interoperable global frameworks. Finally, interoperability of governance approaches is constrained by regulatory fragmentation, with 29 distinct data protection regimes and no unified stance on data sovereignty or infrastructure control. In contrast to more consolidated models such as the EU, this limits Africa's negotiating power. Encouragingly, increasing political momentum, including AU leadership initiatives and continental convenings, indicates readiness for a coordinated approach. The opportunity now lies in leveraging this convergence of political will, investment, and strategic frameworks to consolidate Africa's position within global AI governance, with the Geneva Dialogue serving as a potential catalyst.

The AI Dialogue can play a crucial role in making global AI governance more inclusive, practical, and representative of African priorities. It offers a universal UN forum where African governments, regional bodies, civil society, academia, and industry can participate on a more equal footing in shaping norms on AI safety, accountability, and human rights. For Africa, this matters because the continent is often underrepresented in global technology rule-making, even though AI systems increasingly affect African economies, languages, public services, and labour markets. The Dialogue can strengthen international cooperation in three ways. First, it can support knowledge exchange and technology transfer, helping African institutions access regulatory expertise, evaluation tools, and technical capacity needed to govern AI effectively. Second, it can help align global principles with regional realities by ensuring that issues such as data scarcity, multilingual inclusion, digital infrastructure gaps, and the risks of imported systems are reflected in global discussions. Third, it can promote coordinated responses to cross-border risks, including misinformation, cyber harms, surveillance, and military uses of AI, which cannot be handled by one country alone. For Africa, the Dialogue should not only be consultative but also action-oriented. It should create space for regional blocs such as the African Union to shape interoperable governance approaches, while supporting national AI strategies and capacity-building. If designed well, the Dialogue can help turn Africa from a policy taker into a policy shaper in global AI governance

Several initiatives already provide a foundation for AI governance in Africa, some of which I had already mentioned. The African Union adopted its Continental AI Strategy in 2024, giving member states a shared framework for ethical, responsible, and development-oriented AI. In 2025, African countries also advanced commitments through the AI for Africa Initiative, which emphasises research, capacity-building, investment, and interoperable governance approaches. More recently, Africa-specific policy work has highlighted the creation of an Africa AI Council and stronger coordination mechanisms to improve continental governance. At the global level, the UN's Global Dialogue on AI Governance will begin in July 2026, creating a formal multistakeholder space for coordination. The AI Dialogue's added value is that it can connect these efforts rather than duplicate them. It can serve as a bridge between African regional priorities and global rule-making, ensuring that African perspectives shape emerging standards on safety, transparency, and accountability. It can also help harmonize fragmented initiatives by encouraging interoperability across legal frameworks, sharing best practices, and supporting joint assessments of AI risks and impacts. This is important because African experts have noted that the challenge is often not a lack of rules, but weak coordination and uneven enforcement. In practical terms, the Dialogue could help mobilise resources, promote south-south cooperation, and translate principles into implementation support for regulators and public institutions. For Africa, that means the Dialogue can amplify existing work, reduce duplication, and strengthen the continent's influence in shaping a fair global AI governance architecture

Different stakeholders can contribute by bringing their distinct expertise, lived experience, and practical constraints into the dialogue. Policymakers can clarify regulatory priorities and public-interest goals, industry can share implementation realities and innovation pathways, academia can provide evidence-based analysis, and civil society can raise social, ethical, and rights-based concerns. To support meaningful participation, the AI Dialogue should use a balanced format with short expert inputs, structured breakout discussions, and clear questions focused on decision-relevant issues. Materials should be shared in advance, jargon should be minimised, and inputs should be documented transparently so participants can see how their contributions shape the outcomes.

Underrepresented voices often include communities from the Global South, small and medium-sized enterprises, workers affected by automation, Indigenous communities, people with disabilities, youth, older adults, and groups with limited digital access. These perspectives can also include frontline practitioners, local regulators, consumer advocates, and people directly impacted by AI systems in sectors such as health, education, recruitment, and public services. Inclusion can be improved through targeted outreach, travel and accessibility support, multilingual participation, hybrid or low-bandwidth engagement options, and partnerships with trusted local organisations. It is also important to move beyond symbolic representation and ensure these voices have real influence in agenda-setting, deliberation, and follow-up.

Effective formats could include facilitated roundtables, scenario-based workshops, live polling, and small-group deliberations that focus on specific use cases or governance dilemmas. A "world café" model or rotating breakout sessions can help participants engage with multiple topics while keeping discussions interactive and inclusive. Case-based simulations, stakeholder mapping exercises, and co-design sessions can make the dialogue more practical and help translate abstract principles into actionable recommendations. The most effective format will likely combine plenary sessions for shared framing with smaller, structured sessions for deeper exchange and concrete outputs.

Policy approaches should be risk-based, human-centric, and interoperable across sectors and borders. Good practice includes requiring AI impact assessments before deployment, especially for high-risk uses; assigning clear human accountability for every system; and maintaining traceability through documentation, testing, and incident reporting. The UN's Global Dialogue on AI Governance is explicitly aimed at inclusive, multi-stakeholder deliberation on pressing AI challenges, and current global guidance emphasises managing AI risks, distributing benefits more equitably, and aligning governance rules across jurisdictions. Concrete solutions that work in practice include: Risk-tiered governance, where stronger controls apply to higher-stakes uses such as hiring, credit, health, and public services. Mandatory transparency measures, including model cards, data provenance records, and notices when AI is used in decisions affecting people. Independent evaluations and red-teaming before deployment, plus continuous monitoring after launch. Procurement rules that require vendors to meet security, privacy, and auditability standards. Mechanisms for complaints, contestability, and human review so that affected people can challenge harmful outcomes. A strong policy environment also invests in AI fluency, public-sector capacity, and cross-border coordination so that governance keeps pace with fast-moving systems and agentic AI risks