University of Lubumbashi, DRC

The first Global Dialogue on AI Governance would be successful if it achieves the following outcomes: 1. Concrete Framework for International Cooperation: Success means moving beyond abstract principles to establish actionable mechanisms for cross-border collaboration on AI safety, standards, and risk assessment. This includes agreement on shared terminology, risk classification systems, and protocols for information sharing during AI incidents. 2. Inclusive Representation and Voice: The Dialogue must ensure meaningful participation from the Global South, small nations, and underrepresented communities, not just major AI powers. Success means their concerns about AI colonialism, resource inequality, and cultural values are reflected in governance frameworks, not sidelined. 3. Bridging Technical and Policy Communities: A successful outcome would create permanent channels connecting AI researchers, ethicists, policymakers, and affected communities. This includes establishing working groups that translate technical AI developments into policy-relevant insights and vice versa. 4. Transparency Commitments: Governments and major AI developers should commit to transparency measures: publishing safety evaluations, sharing incident reports, and disclosing training data sources and model capabilities. Even voluntary commitments would signal progress. 5. Addressing Immediate Harms: While discussing future risks, success requires acknowledging current AI harms, algorithmic discrimination, labor displacement, misinformation, and surveillance. Concrete action plans to mitigate these existing issues would demonstrate credibility. 6. Follow-up Mechanisms: The Dialogue should establish clear next steps: regular convenings, accountability structures, and measurable goals. Without implementation roadmaps and progress tracking, even excellent discussions remain theoretical. Ultimately, success means participants leave Geneva with shared understanding, mutual trust, and commitment to collaborative governance, transforming AI from a competitive race into a coordinated effort to benefit humanity while managing risks responsibly.

Safe, secure and trustworthy AI;AI capacity-building;Social, economic, ethical, cultural, linguistic and technical implications of AI;

All four thematic areas are interconnected and equally critical, but priorities 1, 3, and 4 demand the most urgent coordinated action: Priority 1 (Safe, secure and trustworthy AI) is foundational. Without establishing safety standards and security protocols now, we risk deploying increasingly powerful systems with inadequate safeguards. The rapid pace of AI development makes this time-sensitive, governance frameworks must keep pace with technological advancement. Priority 3 (Social, economic, ethical, cultural implications) requires immediate attention because AI is already reshaping labor markets, amplifying biases, and affecting cultural representation. These impacts are not hypothetical futures-they're current realities demanding urgent intervention. Addressing economic displacement, algorithmic discrimination, and cultural erasure cannot wait. Priority 4 (Transparency, accountability, and human oversight) is essential for implementing the other priorities. Without transparency into AI systems' capabilities and decision-making processes, we cannot assess safety, address harms, or ensure accountability. Establishing oversight mechanisms now prevents entrenchment of opaque, unaccountable systems. Priority 2 (AI capacity-building) is crucial for equitable governance but operates on a longer timeline. While urgent, capacity-building requires sustained investment rather than immediate crisis response. The selection reflects the need to simultaneously establish safety guardrails, address ongoing harms, and create accountability structures, all while ensuring these frameworks are developed inclusively through capacity-building efforts.

Several critical issues transcend the listed themes: 1. Environmental Impact and Sustainability: AI's massive energy consumption, water usage for data center cooling, and carbon footprint are absent from the themes. As AI scales, environmental costs grow exponentially. Governance must address sustainable AI development, renewable energy requirements, and climate justice implications. 2. AI and Democratic Integrity: The intersection of AI with elections, information ecosystems, and civic participation deserves explicit focus. Deepfakes, synthetic media, micro-targeted manipulation, and AI-generated disinformation threaten democratic processes globally. This requires coordinated international response beyond general "trustworthiness." 3. Power Concentration and Market Dynamics: The themes don't address monopolistic control over AI infrastructure, compute resources, and foundational models. A handful of corporations control critical AI capabilities, creating dependencies and power asymmetries. Governance must tackle antitrust concerns, infrastructure sovereignty, and preventing AI colonialism. 4. Dual-Use and Militarization: AI's military applications, autonomous weapons, surveillance systems, cyber capabilities, need explicit governance frameworks. The civilian-military dual-use nature of AI technologies requires international protocols similar to nuclear or biological weapons governance. 5. Intergenerational Justice: Current AI decisions create long-term path dependencies affecting future generations. Governance should consider intergenerational equity, reversibility of AI deployments, and preserving human agency for future societies. 6. Neurotechnology and Human Enhancement: Brain-computer interfaces and AI-enabled cognitive enhancement raise unprecedented questions about human identity, autonomy, and inequality that existing themes don't fully capture. These issues require dedicated attention alongside the four core themes.

Challenges: 1. Regulatory Vacuum and Legal Uncertainty: The absence of comprehensive AI governance frameworks in the DRC and most African nations creates legal ambiguity. As a law professor, I observe that our existing legal systems, designed for pre-digital contexts, struggle to address AI-related issues: algorithmic accountability, data sovereignty, automated decision-making in public services, and cross-border data flows. This vacuum leaves citizens unprotected and creates barriers to responsible AI adoption. 2. Capacity and Resource Constraints: African institutions lack technical expertise, funding, and infrastructure to develop contextually appropriate AI governance. The DRC faces acute challenges: limited digital infrastructure, brain drain of technical talent, and competing development priorities. Without capacity-building support, we risk importing governance models that don't reflect our legal traditions, cultural values, or socioeconomic realities. 3. Data Colonialism and Extractive Practices: Foreign AI companies extract African data, including from the DRC's rich linguistic and cultural diversity, without consent frameworks, benefit-sharing, or local oversight. Governance gaps enable exploitative practices where our data trains models that primarily serve external markets, perpetuating digital dependency. 4. Exclusion from Standard-Setting: Global AI governance discussions often proceed without meaningful African participation. Standards developed elsewhere may not address our priorities: agricultural AI for food security, healthcare AI for resource-limited settings, or governance models compatible with customary law systems. Opportunities: 1. Leapfrogging to Rights-Based Frameworks: The DRC and Africa can develop AI governance grounded in human rights, Ubuntu philosophy, and community-centered approaches, offering alternative models to purely market-driven or surveillance-oriented frameworks. 2. Regional Harmonization: African Union initiatives like the Continental AI Strategy provide opportunities for coordinated governance, avoiding fragmentation while respecting national sovereignty. 3. Legal Innovation: African legal scholars can pioneer governance addressing unique contexts: multilingual AI, oral tradition preservation, and AI for sustainable development—contributing distinctive perspectives to global discourse.

The AI Dialogue can play several transformative roles: 1. Inclusive Multilateral Platform: Unlike industry-led initiatives or exclusive clubs of AI-advanced nations, the UN-convened Dialogue provides legitimacy and universality. It can ensure African voices—including from the DRC—are heard equally alongside major AI powers, making cooperation genuinely global rather than imposed. 2. Bridging Fragmented Governance Efforts: AI governance is currently scattered across regional bodies, industry consortia, and bilateral agreements. The Dialogue can serve as a coordination mechanism, identifying overlaps, harmonizing approaches, and preventing contradictory standards that create compliance burdens, especially for developing nations. 3. Norm-Setting and Soft Law Development: The Dialogue can establish shared principles and voluntary frameworks that guide national legislation. For countries like the DRC developing AI laws, internationally recognized norms provide valuable reference points while respecting sovereignty. 4. Capacity-Building Coordination: By connecting countries needing AI governance expertise with those offering technical assistance, the Dialogue can facilitate knowledge transfer, training programs, and institutional support, addressing the capacity gaps hindering African participation. 5. Early Warning and Risk Sharing: The Dialogue can create mechanisms for sharing information about AI incidents, emerging risks, and governance failures. This collective learning benefits all nations, particularly those with limited resources for independent AI monitoring. 6. Accountability Without Enforcement: While lacking binding authority, the Dialogue can establish peer review mechanisms, transparency commitments, and reputational incentives that encourage responsible AI development, complementing rather than replacing national sovereignty. 7. Amplifying Underrepresented Perspectives: The Dialogue can systematically incorporate perspectives from civil society, academia, indigenous communities, and the Global South, ensuring cooperation reflects diverse values, not just technological or economic interests. For the DRC and Africa, the Dialogue represents rare opportunity to shape global AI governance from inception rather than adapting to externally imposed frameworks.

Initiatives to Connect With: 1. African Union Continental AI Strategy: The AU's framework addresses AI governance from African perspectives. The Dialogue should formally engage with AU mechanisms, ensuring continental priorities inform global cooperation and avoiding parallel processes that fragment African participation. 2. UNESCO Recommendation on AI Ethics: As the first global standard on AI ethics, UNESCO's framework provides foundational principles. The Dialogue should operationalize these principles through implementation guidance, monitoring mechanisms, and capacity support, moving from aspiration to action. 3. OECD AI Principles: While valuable, OECD principles reflect primarily developed-nation perspectives. The Dialogue can complement them by incorporating Global South priorities: AI for development, digital sovereignty, and addressing historical inequalities. 4. Partnership on AI (PAI) and Multi-Stakeholder Initiatives Industry-led efforts like PAI offer technical expertise but lack governmental legitimacy. The Dialogue can bridge this gap, translating technical insights into policy frameworks while ensuring corporate accountability. 5. Regional Frameworks (EU AI Act, ASEAN AI Governance): Regional approaches offer lessons in balancing harmonization with local context. The Dialogue can facilitate cross-regional learning and interoperability without imposing one-size-fits-all solutions. 6. Global Partnership on AI (GPAI): GPAI's working groups produce valuable research. The Dialogue should leverage these outputs while expanding membership beyond current participants to include underrepresented regions. Added Value of the AI Dialogue: Universal legitimacy through UN convening power Binding developing and developed nations in shared governance conversations Translating technical discussions into accessible policy language for non-specialist governments Creating implementation pathways for existing principles Establishing accountability mechanisms beyond voluntary commitments Centering human rights and development alongside innovation and security Providing neutral space free from geopolitical bloc dynamics or corporate influence The Dialogue's unique value lies in combining inclusivity, legitimacy, and action-orientation, transforming fragmented AI governance into coordinated international cooperation.

Stakeholder Contributions: - Governments: Provide national perspectives on regulatory challenges, share policy experiments, and commit to implementation of agreed principles. - Civil Society: Represent affected communities, document AI harms, advocate for rights-based approaches, and ensure accountability mechanisms. - Academia: Offer independent research, technical expertise, ethical frameworks, and evidence-based policy recommendations, particularly from legal, social science, and humanities perspectives. - Technical Experts: Translate complex AI capabilities into policy-relevant insights, assess feasibility of proposed governance measures, and identify emerging risks. - Private Sector: Share implementation experiences, explain technical constraints, commit to transparency standards, and engage constructively rather than lobbying against regulation. - Affected Communities: Provide lived experiences of AI impacts, particularly marginalized groups facing algorithmic discrimination, labor displacement, or surveillance. Format and Structure Recommendations: 1. Hybrid Participation: Combine in-person Geneva sessions with robust virtual participation to enable Global South engagement despite travel/visa barriers. 2. Pre-Dialogue Regional Consultations: Hold preparatory meetings in Africa, Asia, Latin America, and Pacific regions to consolidate regional priorities and build capacity for effective participation. 3. Thematic Working Groups: Organize parallel sessions on each priority area, allowing stakeholders to contribute specialized expertise while maintaining plenary sessions for cross-cutting issues. 4. Balanced Panels: Ensure every panel includes voices from multiple regions, sectors, and perspectives—avoiding dominance by AI-advanced nations or industry. 5. Written Submissions Process: Maintain open calls for written inputs (like this one) to include voices unable to attend physically. 6. Translation and Accessibility: Provide real-time interpretation in UN languages plus key regional languages; ensure materials are accessible to persons with disabilities. 7. Follow-Up Mechanism: Establish clear processes for ongoing stakeholder engagement beyond the July event, including implementation monitoring and annual reviews.

Currently Underrepresented: 1. African Nations and Institutions Despite representing 1.4 billion people, African voices are marginalized in AI governance. Inclusion requires: funding for African delegates' participation, partnerships with African universities and civil society, and recognizing African Union frameworks as authoritative regional input. 2. Small Island Developing States (SIDS) Pacific, Caribbean, and Indian Ocean nations face unique AI challenges (climate adaptation, digital sovereignty) but lack resources for participation. Virtual engagement, regional representation mechanisms, and targeted capacity-building are essential. 3. Indigenous Communities AI systems often violate indigenous data sovereignty, cultural protocols, and traditional knowledge. Inclusion requires: recognizing indigenous governance structures, ensuring free prior informed consent, and incorporating indigenous legal principles into frameworks. 4. Workers and Labor Organizations Those experiencing AI-driven displacement, surveillance, and algorithmic management are rarely centered. Trade unions, gig workers' collectives, and labor rights organizations must have formal roles. 5. Persons with Disabilities AI accessibility and assistive technology perspectives are often absent. Disability rights organizations should co-design governance addressing both opportunities and risks. 6. Rural and Low-Connectivity Communities AI governance discussions assume high connectivity and digital literacy. Including rural perspectives requires: offline consultation methods, community radio engagement, and recognizing oral testimony. 7. Legal Scholars from Diverse Traditions AI governance is dominated by Western legal frameworks. Scholars versed in Islamic law, customary law, Ubuntu jurisprudence, and other traditions should contribute alternative governance models. 8. Youth and Future Generations Those who will live longest with AI consequences need representation through youth councils and intergenerational justice frameworks. Inclusion Strategies: Dedicated funding for underrepresented participants Quota systems ensuring regional/sectoral balance Community-based consultation processes Partnerships with grassroots organizations Capacity-building before the Dialogue Recognition of non-English contributions

1. Fishbowl Conversations Inner circle of diverse stakeholders discusses while outer circle observes; participants rotate in/out. This ensures active listening, prevents grandstanding, and allows multiple voices to contribute substantively. 2. Scenario Planning Workshops Small groups develop governance responses to concrete AI scenarios (e.g., agricultural AI failure in food-insecure region, deepfake election interference). This moves beyond abstract principles to practical problem-solving and reveals divergent priorities. 3. Reverse Panels Instead of experts presenting to audiences, affected communities present their experiences while policymakers and technologists listen and respond. This centers lived experience over theoretical expertise. 4. Digital Collaboration Platforms Pre-Dialogue online workspace where stakeholders co-draft proposals, comment on others' submissions, and build coalitions. This extends engagement beyond the two-day event and creates documented consensus-building processes. 5. Lightning Talks and Poster Sessions Allow many voices to share innovations, case studies, and research in short formats, avoiding monopolization by lengthy presentations from well-resourced institutions. 6. Deliberative Polling Survey participants before and after informed deliberation on contentious issues, revealing how perspectives shift with exposure to diverse viewpoints and evidence. 7. Indigenous Talking Circles Incorporate non-Western deliberation formats that emphasize consensus, respect, and holistic thinking—offering alternatives to adversarial debate models. 8. Live Policy Drafting Collaborative sessions where participants draft governance language in real-time, negotiating specific wording and commitments rather than discussing abstractions. 9. Youth-Led Sessions Dedicate segments where young people facilitate discussions, ensuring intergenerational dialogue and fresh perspectives. 10. Accountability Commitments Wall Public space (physical and digital) where governments and organizations post specific, measurable commitments—creating transparency and peer pressure for follow-through. These formats prioritize dialogue over monologue, action over rhetoric, and inclusion over exclusion.

Regulatory Frameworks: 1. EU AI Act's Risk-Based Approach Categorizing AI systems by risk level (unacceptable, high, limited, minimal) provides scalable governance-prohibiting harmful applications while enabling innovation in low-risk areas. This model is adaptable for resource-constrained contexts like the DRC, allowing prioritization of oversight where risks are greatest. 2. Rwanda's AI Policy Framework Rwanda's proactive approach demonstrates that developing nations can lead in AI governance. Their focus on AI for development priorities (agriculture, healthcare, education) while building regulatory capacity offers a model for African countries including the DRC. Sectoral Approaches: 3. Kenya's Data Protection Act Strong data protection legislation provides foundation for AI governance by establishing consent requirements, data subject rights, and oversight mechanisms-essential for preventing exploitative AI practices. Multi-Stakeholder Initiatives: 4. African Observatory on Responsible AI Regional knowledge-sharing platforms that document AI impacts, share governance innovations, and build collective expertise strengthen African capacity for contextually appropriate governance. Community-Centered Approaches: 5. Indigenous Data Sovereignty Principles (CARE, OCAP) Frameworks ensuring indigenous communities control data about themselves offer models for community-based AI governance that respect cultural values and self-determination-applicable beyond indigenous contexts to any community-generated data.