Carnegie Mellon University

I would love the first Global Dialogue on AI Governance should move beyond general principles and produce clear, actionable outcomes that reflect global priorities while remaining practical for implementation. Firstly, success for me would mean establishing a shared baseline for AI governance, including broadly agreed principles for safety, accountability, and human oversight that can guide both national and regional frameworks. I know that full alignment may not be immediately achievable but identifying common ground across Member States and stakeholders would be a critical step toward interoperability. Secondly, I would love that the Dialogue should deliver concrete pathways for bridging AI divides. This should include commitments to capacity building in developing regions, such as access to compute infrastructure, support for open-source AI, and investments in local talent and research ecosystems. Without addressing these gaps, global AI governance risks reinforcing existing inequalities. Thirdly,I also feel that having some good progress on technical and operational safety would be essential. This could include recommendations for evaluation standards, risk assessment practices, and mechanisms for monitoring AI systems in real-world deployment, ensuring that safety is not only policy-driven but also technically grounded is key. Finally, the Dialogue should strengthen multi-stakeholder collaboration, creating structures for ongoing engagement between governments, academia, industry, students and civil society. AI governance is a continuous process, and sustained coordination will be necessary to adapt to rapid technological change. Overall, the Dialogue would be successful if it produces practical frameworks, inclusive commitments, and sustained collaboration mechanisms that can guide responsible AI development globally while reflecting diverse regional contexts. Also including students interested in this space like myself in committee and projects to foster their growth and development

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• AI capacity-building

My priorities reflect the need to ensure that AI systems are both responsibly developed and equitably accessible, particularly within emerging and diverse contexts. Safe, secure and trustworthy AI is a top priority because the rapid deployment of AI systems without robust safety mechanisms introduces significant risks. Ensuring reliability, robustness, and alignment in real-world environments is very key to prevent unintended harm and build public trust. AI capacity-building is equally critical, especially in developing regions (such as Africa). Bridging gaps in access to compute, data, and technical expertise is necessary to ensure that countries can actively participate in AI development rather than remain passive consumers. This includes strengthening local talent, research ecosystems, and infrastructure. The social, economic, ethical, cultural, linguistic, and technical implications of AI are central to ensuring that AI systems are inclusive and context-aware. In many regions, including Africa, there is a need to ensure that AI reflects local realities, languages, and societal values, while addressing risks such as bias, exclusion, and unequal impact. Finally, interoperability of governance approaches is important to avoid fragmented regulatory systems. As AI systems operate across borders, aligning governance frameworks will support collaboration, consistency, and effective oversight globally.

The technical aspect of AI safety and ethics on an organizational level need to be also prioritized. Many of the cyber attacks are targeted mostly to organizations and the use of Agentic systems has become the hot source of entrying. There is need for organization to focus and invest on AI safety and understand how to protect themselves. There is need for a lot of awareness and promotion of technological tools that act has guardrails which the UN can support and promote.

Governance gaps in AI are already visible in both Nigeria and Rwanda, particularly in how rapidly AI tools are being adopted compared to the pace of policy, standards, and technical safeguards. I am talking about Nigeria and Rwanda because I am from Nigeria but currently relocated to Rwanda for a Masters which is focused on Applied ML. In safe, secure, and trustworthy AI, there is increasing use of AI in sectors such as finance, public services, and digital platforms, but limited standardized approaches to model evaluation, risk assessment, and deployment monitoring. This creates exposure to issues such as unreliable outputs, bias, and security vulnerabilities, especially in high-impact systems. For AI capacity-building, both countries are making progress through growing tech ecosystems and academic programs, but there are still gaps in access to advanced compute resources, large-scale datasets, and specialized AI training. This limits the ability to develop and deploy locally relevant AI systems at scale and increases reliance on external technologies. The social, cultural, and linguistic implications are particularly significant. Many AI systems are not optimized for local languages like Yoruba, Hausa, Igbo, or Kinyarwanda, leading to exclusion and reduced effectiveness. This also raises concerns about representation, bias, and the risk of reinforcing global inequalities in AI development. Finally, the lack of interoperable governance frameworks across countries creates fragmentation. As AI systems and digital services operate across borders, inconsistent policies make it difficult to ensure accountability, data protection, and coordinated oversight.

The AI Dialogue can play a critical role as a neutral, multi-stakeholder platform for aligning global efforts on AI governance. One of its most important contributions would be to help establish shared principles and practical frameworks that countries can adapt locally, while still enabling coordination across borders. This is especially important as AI systems are inherently global, but governance remains largely national. It can also tighten cooperation by bridging gaps between regions at different stages of AI development. By bringing together governments, students, academia, industry, and civil society, the Dialogue can support knowledge sharing, capacity-building partnerships, and coordinated investments in areas such as compute infrastructure, open-source AI, and skills development, ensuring that developing regions are active participants in shaping AI, not just adopters. In addition, the Dialogue can promote interoperability of governance approaches, helping reduce fragmentation across regulatory systems. This includes encouraging alignment on safety standards, risk assessment practices, and accountability mechanisms, which are essential for cross-border AI deployment and oversight. We can start with policy frameworks that generically align across some regions such as East Africa, West Africa, etc. Finally, it can serve as a space for continuous collaboration and iteration, where emerging risks, technical developments, and policy responses are regularly discussed. This ongoing engagement is key to ensuring that governance keeps pace with rapid advancements in AI. Above all, the AI Dialogue can strengthen international cooperation by fostering alignment, inclusion, and sustained coordination in how AI is governed globally.

A number of existing initiatives already shape global AI governance, and the AI Dialogue should build on these rather than duplicate efforts. Also earlier in the year, during a session I had with Dr. Chinasa Okolo, who currently works with the United Nations Office for Digital and Emerging Technologies as AI & Emerging Technologies Policy Specialist, when she came to my school (CMU Africa), mentioned some ongoing frameworks being worked on even at the UN level. First, multilateral frameworks such as UNESCO's Recommendation on the Ethics of AI provide widely endorsed principles on human rights, transparency, and accountability. Similarly, the OECD AI Principles and the Global Partnership on Artificial Intelligence offer guidance and collaboration platforms across governments and experts. Regional efforts like the African Union's AI strategy are also critical in reflecting local priorities and contexts. Second, technical and standards-driven bodies like the International Organization for Standardization(IOS) and Institute of Electrical and Electronics Engineers (IEEE) contribute to operationalizing AI governance through standards on safety, risk management, and system design. In parallel, open-source and research communities continue to advance tools for AI evaluation, transparency, and responsible deployment. The UN AI Dialogue also adds value by acting as a coordinating layer across these efforts. Rather than creating new principles, it can help translate existing ones into practical, interoperable approaches, especially for countries with limited implementation capacity. It can also strengthen inclusion, ensuring that perspectives from developing regions, particularly across Africa, are meaningfully integrated into global governance discussions. Additionally, the Dialogue can facilitate linkages between policy and technical practice, connecting high-level frameworks with real-world deployment challenges. By fostering continuous engagement across stakeholders, it can help ensure that governance evolves alongside technological progress.

Different stakeholders should contribute to the AI Dialogue based on their strengths, while working within a structured, outcome-oriented format. Governments can provide policy direction, share national strategies, and commit to implementing agreed principles. Last week, I heard about the National AI policy draft that the Kenya government is working on. Even though it is not perfect, it is a progressive step on the national level. Industry can contribute practical insights on system development, deployment challenges, and safety practices. They can also sponsor research just like Ericsson does with some universities in Europe while allowing them to use their staff for experiments and surveys in addition. Academia and the technical community can offer research-based evidence, evaluation methods, and emerging innovations. They can also facilitate research around this area with strategic partnerships that could fund it and also involve students who can also learn and grow in the process. Civil society can ensure that human rights, inclusion, and societal impacts are fully represented, especially for underrepresented groups. To be effective, the Dialogue should be structured in a way that balances broad participation with focused outcomes. First, I would suggest that we adopt a thematic working group model, aligned with key areas such as safety, capacity-building, and governance interoperability. These groups should include diverse stakeholders and produce clear outputs, such as recommendations, frameworks, or best practices. Secondly, the Dialogue could include technical-policy interface sessions, where engineers and policymakers collaboratively translate high-level principles into implementable guidelines. This helps bridge the gap between policy intent and real-world system design. Thirdly, there should be a regional consultation layer, ensuring that inputs from different regions, especially developing countries are incorporated into global discussions in a structured way. Also I would suggest that the Dialogue should not be a one-time event but an ongoing process

First, stakeholders from developing regions, particularly across Africa, are often underrepresented. This leads to governance models that may not fully reflect local realities, infrastructure constraints, or societal priorities. Closely related are local researchers, engineers, and startups, who are building and deploying AI in these contexts but are rarely included in global decision-making spaces. Second, non-English language communities and cultural groups are often overlooked. This affects how issues such as linguistic inclusion, cultural context, and local data representation are addressed in AI systems. I recently started research on building a realtime speech to speech system for local African languages starting with Rwanda's local language and we are facing this issue. Finally, there is often limited participation from interdisciplinary experts, including social scientists, ethicists, and practitioners working at the intersection of technology and society. To improve inclusion, the AI Dialogue should adopt regionally grounded consultation processes, ensuring that inputs are gathered from diverse local ecosystems before global discussions. It should also support capacity-building and funding mechanisms that enable participation from underrepresented groups, including travel support, research grants, and remote engagement options. Additionally, incorporating multilingual engagement, partnerships with local institutions, and structured pathways for community input can help ensure broader representation. Including these perspectives is essential for building AI governance frameworks that are equitable, context-aware, and globally relevant.

To push meaningful and dynamic engagement, I will suggest that the AI Dialogue should move beyond traditional panel discussions and adopt more interactive, outcome-driven formats. First, setting up co-creation labs or policy sprints that can bring together policymakers, engineers, and researchers to collaboratively design practical solutions, such as draft guidelines, risk frameworks, or implementation roadmaps. There could be technical-policy simulation exercises and systems that can help participants explore real-world scenarios, such as deploying AI in healthcare or finance, and work through governance challenges in a controlled, collaborative setting. This encourages deeper understanding of trade-offs and decision-making under uncertainty. There could be regional roundtables with global integration that can ensure diverse perspectives are captured. Insights from region-specific discussions can be synthesized and fed into global sessions, creating a more inclusive and representative dialogue. Organizing open innovation or challenge tracks where we invite participants to propose solutions to specific governance problems, such as improving AI transparency or bridging capacity gaps, encouraging broader participation and fresh ideas. Overall, the most effective formats will be those that combine collaboration, practical problem-solving, and inclusivity, leading to actionable outcomes rather than just discussion.

OECD AI Principles: The OECD AI Principles, adopted in 2019 and updated with broad international support, are one of the first intergovernmental standards on AI. They are non-binding but highly influential and have been endorsed by over 40 countries.They focus on five core values: Inclusive growth, sustainable development, and well-being, Human-centered values and fairness, Transparency and explainability, Robustness, security, and safety, Accountability OECD AI Policy Observatory (OECD.AI): This is one of the most comprehensive global AI governance platforms and this does the following: Tracks AI policies across countries, Hosts a database of national AI strategies and regulations, Provides tools to compare governance approaches, Shares indicators on AI readiness, adoption, and risk UNESCO Recommendation on the Ethics of Artificial Intelligence: it was adopted in 2021, this is the first global normative framework on AI ethics endorsed by all UNESCO member states. It goes beyond technical governance and strongly emphasizes human rights and dignity. Key principles include: Human rights and fundamental freedoms, Transparency and explainability, Fairness and non-discrimination, Environmental and social well-being, Accountability and oversight. AI Incident Database: A public platform for documenting real-world AI failures and harms. African Union (AU) Continental AI Strategy: This is the most important continent-wide framework and it is a unified strategy adopted in 2024 to guide AI development across African countries.