SEGEL

A successful first Global Dialogue on AI Governance should move beyond broad principles and create practical direction for governments, civil society, academia, the private sector and non-profit organisations to adopt responsible AI governance systems. From my perspective as a development professional and AI enthusiast working in Africa and the Middle East, success would mean that the Dialogue gives meaningful voice to countries and organisations that are often underrepresented in technology governance. One important outcome should be a clear call for organisations to proactively establish AI ethics, governance and risk management systems before AI use becomes informal, fragmented or unmanaged. In many non-profit, humanitarian and development organisations, individual employees are adopting AI tools much faster than their institutions are developing policies, safeguards, skills and accountability mechanisms. This gap creates risks but also shows strong demand for practical, enabling governance. The Dialogue should therefore promote organisational readiness: AI policies, ethical review processes, staff guidance, data protection safeguards, human oversight, and responsible experimentation. It should also advance a new approach to AI capacity strengthening that goes beyond one-off training and supports institutional systems, leadership awareness, digital infrastructure, local language inclusion and peer learning. For countries such as Ethiopia and most African countries and the Global South in general, success would include support for local AI literacy, research partnerships, public-interest use cases, and safeguards against exclusion, bias and misuse. The Dialogue should produce clear follow-up mechanisms, measurable milestones and inclusive stakeholder engagement so that AI governance becomes practical, locally relevant and development-oriented.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight
• Protection and promotion of human rights

I selected these priorities because they are essential for enabling responsible, inclusive and practical AI adoption, particularly in the non-profit, humanitarian and development sectors. Safe, secure and trustworthy AI is foundational because many organisations are already using AI tools, often informally, before internal governance systems are in place. This is especially visible in non-profit organisations, where employees may adopt AI to improve productivity, proposal writing, research, analysis, communication or programme delivery faster than organisations can issue guidance, assess risks or build accountability systems. Governance must therefore be proactive, enabling and proportionate, rather than reactive. AI capacity-building is equally urgent. A new capacity-strengthening approach is needed: one that goes beyond short-term training and builds organisational readiness, leadership awareness, practical policies, ethical review mechanisms, responsible data systems, local technical skills and peer-learning platforms. This is critical for narrowing the digital divide between countries, institutions and communities with different levels of access to AI tools, infrastructure and expertise. Protection and promotion of human rights must remain central because AI can affect access to services, information, livelihoods, education and civic participation. Strong safeguards are needed to prevent discrimination, exclusion, privacy violations and misuse. Transparency, accountability and human oversight are necessary to ensure that AI-supported decisions remain explainable, contestable and aligned with public interest. Together, these four priorities support innovation while ensuring that organisations adopt AI responsibly, ethically and in ways that strengthen rather than weaken trust.

Yes. One major cross-cutting issue is the widening gap between individual AI adoption and organisational AI governance. In many sectors, especially non-profit, humanitarian, development and research organisations, staff members are already using AI tools in their daily work. However, many organisations have not yet developed clear policies, ethical guidance, risk management systems, data protection procedures or accountability mechanisms. This creates a governance lag that should be treated as an urgent institutional risk and capacity issue. A second emerging issue is the need to mainstream AI ethics and governance into organisational systems. Responsible AI should not be treated as a stand-alone technical topic. It should be integrated into programme design, MEAL systems, research ethics, procurement, safeguarding, data protection, human resources, knowledge management, partnerships and leadership decision-making. Third, capacity strengthening needs to be redefined. Training individuals is important, but it is not enough. Organisations need support to build practical AI governance frameworks, internal learning systems, leadership capability, digital infrastructure, local-language tools, and communities of practice. This is particularly important for narrowing the digital divide between the Global North and Global South, and between large international organisations and local civil society actors. Additional cross-cutting issues include data justice, linguistic inclusion, environmental sustainability, and safeguards for AI use in fragile, conflict-affected and humanitarian settings. Local actors should be supported as co-creators of responsible AI solutions, not only as users of externally designed technologies.

Governance gaps in AI are already affecting Ethiopia, the wider Horn of Africa, and the non-profit, humanitarian and development sectors. The most significant challenge is that AI adoption is moving faster than institutional governance. Many professionals are already using AI for proposal development, translation, research, data analysis, reporting, communication and programme design, while many organisations still lack clear AI policies, ethical guidance, risk assessment tools, data protection safeguards and accountability mechanisms. This creates risks around confidentiality, misuse of sensitive community data, bias, misinformation, weak human oversight and unequal access to AI benefits. In humanitarian and development contexts, these risks are especially serious because organisations often work with vulnerable groups, including displaced people, children, women, people affected by conflict, and communities facing poverty or exclusion. Weak governance can also affect trust between organisations, communities, governments and donors. At the same time, AI presents major opportunities. If governed well, AI can strengthen programme quality, evidence generation, needs assessment, early warning, monitoring and evaluation, knowledge management, translation, inclusive communication and adaptive learning. It can help local organisations improve efficiency and compete more effectively in research, consultancy, humanitarian and development work. The key opportunity is to move from fragmented individual use to responsible organisational adoption. This requires mainstreaming AI ethics and governance into organisational systems, including MEAL, research ethics, safeguarding, data protection, procurement, human resources and leadership decision-making. It also requires a new capacity-strengthening approach that supports institutions, not only individuals, through practical policies, leadership awareness, digital infrastructure, local-language tools and peer learning. For Ethiopia and the region, responsible AI governance can help narrow the digital divide, strengthen local innovation, and ensure that AI supports human rights, accountability and inclusive development rather than deepening existing inequalities.

The AI Dialogue can play a critical role as a trusted global platform for aligning principles, sharing practical experience, and mobilising cooperation on responsible AI governance. Its most important contribution should be to connect global norms with national, regional and organisational implementation. From the perspective of Ethiopia, the Africa and the non-profit/development sector, the Dialogue should help ensure that AI governance is not shaped only by technologically advanced countries and large private actors. It should create space for low- and middle-income countries, local civil society, academia, humanitarian actors, development organisations, youth, women, and communities affected by digital exclusion to influence the global agenda. The Dialogue can also promote cooperation around capacity strengthening. This should include support for AI literacy, institutional readiness, ethical governance systems, data protection, local-language tools, research partnerships, and responsible public-interest AI applications. A major priority should be helping organisations proactively adopt AI ethics and governance frameworks before employee-level AI use becomes unmanaged or risky. In addition, the Dialogue can encourage interoperability between governance approaches while respecting different legal, cultural, economic and development contexts. It can support shared guidance on transparency, accountability, human oversight, human rights protection, and risk management. Its added value would be strongest if it produces practical follow-up mechanisms: communities of practice, technical support platforms, policy toolkits, peer learning, financing pathways, and measurable commitments. In this way, the AI Dialogue can help transform international cooperation from high-level discussion into concrete action that narrows the digital divide and enables safe, inclusive and development-oriented AI adoption.

The AI Dialogue should build upon existing global, regional and sectoral initiatives rather than duplicate them. Relevant mechanisms include UNESCO's work on the ethics of AI, the Global Digital Compact, the Internet Governance Forum, OECD AI principles, African Union digital transformation and AI policy processes, regional economic community initiatives, national digital transformation strategies, data protection frameworks, university research networks, and civil society platforms working on digital rights and responsible technology. It should also connect with humanitarian and development coordination mechanisms, including UN agencies, international NGOs, local NGOs, donor coordination platforms, research institutions, and professional communities working on MEAL, accountability, safeguarding, data responsibility and community engagement. These sectors already manage sensitive information and work with vulnerable populations, so their experience is highly relevant for practical AI governance. The Dialogue should also engage technology companies, open-source communities, academic institutions and local innovation hubs, especially those working on low-resource languages, inclusive digital tools, public-service delivery, climate resilience, education, health, and humanitarian response. The added value of the AI Dialogue would be to bring these fragmented efforts into a more coherent global learning and cooperation space. It can help translate broad principles into practical organisational guidance, especially for governments, civil society and non-profit organisations that lack internal AI governance systems. It can also promote peer learning between countries and sectors, support locally relevant capacity strengthening, and create pathways for resource mobilisation. Most importantly, the Dialogue can help ensure that AI governance is inclusive, rights-based and development-oriented. It should amplify Global South priorities, support local innovation ecosystems, and encourage organisations to mainstream AI ethics, accountability and human oversight into their existing systems before risks become institutionalised.

Different stakeholders should contribute according to their roles and comparative strengths. Governments can share regulatory experiences, national priorities and public-interest safeguards. UN agencies and regional bodies can support coordination, standards alignment and capacity strengthening. Civil society and community-based organisations can bring perspectives on rights, inclusion, accountability and lived experience. Academia can contribute independent research, ethics frameworks and evidence on social impacts. The private sector and open-source communities can share technical expertise, innovation pathways and practical risk mitigation tools. Non-profit and humanitarian organisations can provide lessons on responsible AI use in sensitive development and crisis contexts. The AI Dialogue should be structured as both a high-level policy forum and a practical learning platform. It should include plenary sessions for political direction, thematic tracks for technical discussion, and regional or sector-specific sessions to ensure contextual relevance. Dedicated sessions should focus on organisational AI governance, especially for non-profit, humanitarian and public-interest institutions where employee adoption is moving faster than institutional guidance. The Dialogue should also include pre-dialogue consultations, multilingual participation, youth and civil society forums, and mechanisms for written submissions from local organisations that may not be able to attend physically. Outputs should include practical toolkits, policy guidance, capacity-strengthening commitments, communities of practice and clear follow-up mechanisms. To be effective, the structure should avoid one-way speeches only. It should prioritise problem-solving, peer exchange, case-based learning and co-creation of practical governance tools that can be adapted by countries, sectors and organisations.

Several voices remain underrepresented in global AI governance discussions. These include low- and middle-income countries, local civil society organisations, grassroots community groups, women's rights organisations, youth, persons with disabilities, displaced people, indigenous communities, linguistic minorities, and communities living in fragile, conflict-affected or humanitarian settings. Local non-profit and community-based organisations are especially underrepresented, despite being close to the people most affected by development, humanitarian and digital inclusion challenges. Their perspectives are essential because they understand local risks, cultural contexts, languages, institutional capacity gaps and community trust issues. African researchers, universities, local technology innovators, data protection practitioners, evaluators, social scientists and public-sector implementers should also have stronger representation. AI governance is often treated as a technical or legal issue, but it is also a social, ethical, institutional and development issue. Voices from MEAL, safeguarding, education, health, agriculture, climate resilience, urban planning and humanitarian response can help make the discussion more grounded. These groups can be included through funded participation, regional consultations, hybrid access, multilingual interpretation, simplified submission processes, local-language engagement, youth fellowships, civil society advisory groups and partnerships with local universities and networks. The Dialogue should also create safe spaces for affected communities to speak about risks such as exclusion, surveillance, bias, misinformation and lack of accountability. Inclusion should not be symbolic. Underrepresented groups should help shape agendas, define priorities, review outputs and participate in follow-up mechanisms. This will make AI governance more legitimate, practical and responsive to real development contexts.

The AI Dialogue should use interactive formats that move beyond formal speeches and allow participants to co-create practical solutions. One useful format would be regional and sectoral "governance labs," where governments, civil society, academia, private sector actors and development organisations jointly examine real AI use cases and identify risks, safeguards and implementation needs. Another effective format would be organisational AI readiness clinics. These would help non-profit, humanitarian, public-sector and local civil society organisations assess their current AI use, identify governance gaps, and develop practical next steps such as AI policies, staff guidance, data protection measures, ethical review processes and human oversight mechanisms. Scenario-based exercises would also be valuable. Participants could discuss realistic cases involving AI in humanitarian response, education, health, translation, public services, monitoring and evaluation, or misinformation management. This would help translate principles into operational decisions. The Dialogue could also host multilingual digital consultations, youth innovation forums, community listening sessions, policy hackathons, peer-learning circles, and South-South exchange sessions. These formats would allow more diverse voices to contribute, including those unable to attend in person. A living knowledge platform could capture tools, case studies, lessons, policy templates and capacity-strengthening resources. This would make the Dialogue useful beyond the event itself. Finally, the Dialogue should include commitment sessions where stakeholders announce concrete actions, such as developing organisational AI governance frameworks, funding capacity-building, supporting local-language AI tools, or creating regional communities of practice. This would make engagement dynamic, accountable and implementation-oriented.

Effective AI governance can be promoted through a combination of organisational policies, practical tools, participatory processes and capacity-strengthening platforms. First, organisations should adopt internal AI ethics and governance policies. These should define acceptable and prohibited uses of AI, data protection requirements, human oversight responsibilities, transparency standards, and procedures for reviewing higher-risk AI applications. In the non-profit and development sector, such policies should be integrated into MEAL, research ethics, safeguarding, procurement, HR, communications and programme quality systems. Second, organisations can use AI risk assessment and impact assessment tools before deploying AI in sensitive areas. These tools should examine risks related to bias, privacy, exclusion, misinformation, human rights, accountability and potential harm to vulnerable groups. They should also require mitigation measures and clear responsibility for decisions supported by AI. Third, responsible data governance practices are essential. These include data minimisation, informed consent, secure storage, anonymisation where appropriate, community accountability, and careful management of sensitive data. This is especially important in humanitarian, research and development contexts. Fourth, practical AI capacity-strengthening platforms are needed. These could include AI readiness clinics, peer-learning communities, helpdesks, open training resources, and regional centres of excellence that support governments, civil society and local organisations to develop context-appropriate governance systems. Fifth, participatory governance approaches should be promoted. Communities, civil society, affected groups, researchers and frontline workers should be involved in assessing AI risks and shaping safeguards. This is important for trust, inclusion and local relevance. Finally, open and inclusive AI ecosystems can support better governance when combined with safeguards. Open-source tools, local-language datasets, public-interest AI sandboxes, university partnerships and South-South learning platforms can help narrow the digital divide while encouraging transparency, innovation and accountability. Together, these approaches can move AI governance from abstract principles to practical organisational and institutional systems.