Permanent Mission of the United Arab Emirates to the United Nations

A successful first Global Dialogue should deliver concrete, actionable outcomes that move beyond declaratory statements and translate the mandate of GA Resolution 79/325 into measurable progress. The UAE views success along four dimensions: First, advancing AI for Development as a cross-cutting priority. Given the urgent need to accelerate action on the 2030 Agenda, the Dialogue should produce tangible commitments to deploy AI in service of the SDGs. This should include addressing both the risks and opportunities of the environmental impacts of AI. Second, narrowing AI divides. Success means raising global ambitions for universal connectivity, supporting AI infrastructure development, and addressing AI divide through promoting capacity building and joint, transparent, responsible, scientific research, knowledge sharing and the dissemination of best practices, with safeguards in place, to enhance technological competencies. Third, operationalizing safety and risk-sharing. For example, the Dialogue could facilitate voluntary mechanisms, consistent with national laws, regulations and applicable international obligations, for AI companies to share appropriate and non-sensitive safety testing information, where feasible, in advance of model release. The Dialogue should also support voluntary coordination among member states on AI-related incidents, with due regard to national policies and differing national capacities, to ensure that countries outside the small group developing frontier models are not the last to learn of systemic risks. Fourth, anchoring human rights—particularly women's and children's rights—in the governance architecture. A successful Dialogue would generate non-binding but actionable outcomes on gender-responsive AI auditing and an international knowledge-sharing platform on child digital safety and AI based on each country national priority. Ultimately, success means establishing the Dialogue as the central, inclusive forum where capacity-building, infrastructure, safety, rights protection, and development solutions are advanced together rather than in silos.

• Safe, secure and trustworthy AI
• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights

These four priorities reflect the UAE's conviction that AI governance must simultaneously expand opportunity, manage risk, and protect people. On the social, economic, ethical, cultural and technical implications of AI, the UAE prioritizes AI for Development and the environmental, water, and energy implications of AI, as well as deployment of AI to enhance humanitarian aid systems. Experiences such as our Thimphu AI for Development workshop demonstrate how LLM chatbots, summarizers, and matching applications can advance national priorities in tourism, agriculture, and entrepreneurship across the Global South. At the same time, power and water are emerging as binding constraints on AI uptake, particularly in water-stressed markets where infrastructure decisions intersect with economic resilience and national security. On AI capacity-building, the UAE emphasizes the urgent need to address gaps in ICT infrastructure, compute and data center capacity, as well as promote capacity building and joint research and knowledge sharing, with safeguards in place, to enhance technological competencies. On safe, secure and trustworthy AI, the UAE prioritizes responsible use of AI and the protection and security of AI infrastructure itself, recognizing that infrastructure security is foundational to trust. On the protection and promotion of human rights, the UAE prioritizes women's and children's rights, addressing documented systemic harms including algorithmic discrimination against women, technology-facilitated gender-based violence (TFGBV), and child safety risks in AI-powered platforms. ILO data from 2026 confirm that women face disproportionately higher exposure to workplace displacement from generative AI, underscoring the urgency of gender-responsive governance. Another priority is the rights and accessibility for persons with disabilities in the context of AI.

Several cross-cutting issues warrant explicit attention even though they straddle the listed themes: AI infrastructure as a development and security issue. The listed themes treat capacity-building and safety separately, but AI infrastructure-data centers, compute, fiber, energy supply-sits at their intersection. AI infrastructure has fast become essential for effective delivery of many public services: as well as considering the safety risks for those engaging with AI, the protection of AI infrastructure should be in consideration. Environmental and resource constraints on AI. The sustainability of data centers has emerged as a critical consideration in the global AI infrastructure conversation. As energy demands from AI workloads continue to grow, attention has turned to how data centers can be powered more responsibly including through renewable energy sources, nuclear power and gas alongside innovations in cooling technologies that reduce freshwater consumption. Technology-facilitated gender-based violence (TFGBV). AI risks to women and girls are a fast-growing rights violation that current governance frameworks address inadequately. This sits across "human rights," "transparency and accountability," and "safe and trustworthy AI"-and risks falling between the cracks if treated as belonging to any one cluster alone. Child safety in AI environments. The absence of internationally agreed standards for child-safe AI design creates regulatory fragmentation and governance gaps globally. Children's exposure to harmful content, profiling, and exploitation through AI-powered platforms is a distinctive issue requiring dedicated treatment. Structural exclusion from AI governance itself. The underrepresentation of women, youth, and persons with disabilities in AI development, standards-setting, and regulatory processes is itself a human rights concern-and a determinant of whether the other themes will be addressed inclusively.

Challenges: The most acute challenge is the widening AI divide. Much of the world faces gaps in fiber-optic infrastructure, cloud access, and energy reliability, with affordability barriers persisting even where infrastructure exists. Critically, because frontier models are developed in only a handful of countries by a handful of companies, the rest of the world is often last to receive information about AI-related risks—leaving developing states unprepared for systemic harms. For water-stressed regions, including parts of the Gulf, AI infrastructure expansion creates direct tension with water supply management, which policymakers increasingly view through an economic resilience and national security lens rather than as a routine utility issue. On human rights, governance gaps allow algorithmic discrimination against women in recruitment, credit, healthcare, and public services to persist. ILO data from 2026 confirm disproportionately higher exposure of women to workplace displacement from generative AI. TFGBV and child safety risks remain under-addressed globally. Opportunities: AI offers transformative potential for sustainable development. The UAE-Bhutan Thimphu workshop showed how AI-powered tourism platforms, mentor-matching tools for entrepreneurs, and satellite-imagery-driven crop diagnostics can advance national priorities while respecting local values such as Bhutan's Gross National Happiness framework. Technology partnerships between developing states and AI-leading countries and firms can lower investment barriers and enhance domestic capabilities without overtaxing local resources. AI also strengthens environmental management—tracking deforestation, mapping ecosystems, and monitoring environmental damage—provided that human expert judgment remains central. The UAE's own initiatives, including the Federal Decree-Law No. 26 of 2025 on Child Digital Safety, demonstrate how national frameworks can model good practice internationally.

The AI Dialogue can play a uniquely valuable role precisely because no existing forum brings together the universal membership, normative authority, and multi-stakeholder reach required to address AI governance comprehensively. First, the Dialogue can serve as a convening platform that bridges the AI "haves" and "have-nots." With leading models developed in only a handful of countries, the Dialogue can ensure that information about AI-related risks and emerging best practices reaches all member states in a timely fashion—rather than reaching the Global South last. Second, it can enhance platforms to promote capacity building. The Dialogue can discuss establishing improved international frameworks to address AI divide through promoting capacity building and joint, transparent, responsible, scientific research, knowledge sharing and the dissemination of best practices, with safeguards in place, to enhance technological competencies. Third, it can build consensus on standards in areas where fragmentation is most damaging—particularly child-safe AI design and gender-responsive impact assessments. International consensus on minimum safety standards for AI accessible to minors would directly reduce governance fragmentation. Fourth, it can drive cooperation on AI safety and incident response. The Dialogue can help operationalize mechanisms for AI companies to share safety testing results before model release, and for member states to coordinate on AI-related incidents. Fifth, it can mainstream AI for Development across UN bodies, setting new standards for UN development agencies on optimizing projects through AI access and use and telecom and digital tools. Consideration should also be given to applying AI within UN humanitarian agencies.

The Dialogue should build upon and connect with several existing initiatives rather than duplicating them: AI for Development partnerships. The July 2025 AI for Development workshop convened in Thimphu by the Royal Government of Bhutan and the UAE Ministry of Foreign Affairs offers a replicable model of South-South and triangular cooperation, demonstrating how AI tools can be tailored to national development priorities such as tourism, agriculture, and entrepreneurship. National frameworks on child digital safety. The UAE's Federal Decree-Law No. 26 of 2025 on Child Digital Safety, and the associated UAE Child Digital Safety Council, provide a potential model for international knowledge-sharing on regulatory approaches, enforcement experiences, and technical standards for protecting children from AI-related harms. The Independent International Scientific Panel on AI. The Dialogue can ensure that the Panel's annual report integrates dedicated metrics tracking AI's impact, for example on issues such as gender equality and child wellbeing, providing regular, evidence-based accountability against human rights commitments. Multi-stakeholder safety testing efforts. Existing voluntary commitments by leading AI companies on safety testing should be expanded into a more systematic mechanism. Added value of the Dialogue: universal legitimacy, the ability to bring developing-country perspectives to the center of agenda-setting, the convening power to address access to compute and AI models, and the normative authority to embed human rights—including women's and children's rights—into the architecture of AI governance.

The Dialogue's value will depend on whether it brings together not only states but also AI developers, civil society, technical communities, and affected populations in a structured way. Recommended format: • Thematic clusters mirroring Resolution 79/325, with dedicated sub-tracks for cross-cutting issues such as AI for Development, AI infrastructure, and women's and children's rights. • Multi-stakeholder working groups between plenary sessions, tasked with producing concrete deliverables—for example, a voluntary initiative on gender-responsive AI auditing, or a knowledge-sharing platform on child digital safety. • Regional preparatory sessions to ensure that Global South perspectives, including those of small developing states, shape the agenda before it is finalized. • Standing channels for industry input on safety testing, model access, and infrastructure, balanced with civil society input. Stakeholder contributions: • States can offer national models (e.g., the UAE's Child Digital Safety Council and Federal Decree-Law No. 26 of 2025), co-sponsor capacity-building workshops, and commit to non-binding actionable outcomes such as voluntary initiatives on gender-responsive AI auditing. • AI companies can commit to sharing safety testing results in advance of model release, enabling all governments to prepare for new risks. • MDBs and philanthropic partners can co-fund shared compute and connectivity initiatives that fill commercially unviable gaps. • Civil society and technical communities can contribute methodologies for gender-disaggregated algorithmic impact assessments and child-safe AI design. • The Independent International Scientific Panel on AI can provide ongoing evidence and metrics, including on AI's impact on gender equality and child wellbeing. The structure should privilege deliverables over declarations, with each Dialogue cycle expected to produce tangible outputs that build cumulatively over time.

Several constituencies are systematically underrepresented in current global AI governance discussions: Small developing countries and Global South states. Much of the conversation on frontier AI is concentrated in a handful of capitals where leading models are developed. Smaller developing states often lack the technical capacity to engage at the depth required, and their priorities—agriculture, tourism, entrepreneurship, basic connectivity—are often underweighted. The Thimphu AI for Development workshop, co-convened by Bhutan and the UAE, illustrates how targeted, regionally grounded engagement can elevate these perspectives. Women. The structural exclusion of women from AI design, standards-setting bodies, and regulatory processes is itself a human rights concern, and it compounds risks of algorithmic discrimination in recruitment, credit scoring, healthcare, and public services. ILO data from 2026 confirm that women face disproportionately higher exposure to workplace displacement from generative AI, yet they remain underrepresented in shaping governance responses. The Dialogue should specifically include communities who have been the victims of the abuses of AI. Persons with Disabilities. Similar structural exclusion concerns equally apply to persons with disabilities. Children and youth. Despite being among the most affected by AI-powered platforms—through exposure to harmful content, profiling, and exploitation—youth and children's perspectives are rarely embedded in governance design. The absence of internationally agreed standards for child-safe AI design reflects this gap. Recommendations for inclusion: • Regional capacity-building partnerships, building on the UAE's South-South and triangular cooperation model, to prepare developing-country governments to engage substantively based on sharing best practices and lessons learnt. • Mandated gender-disaggregated representation on Dialogue panels and working groups, complementing gender-disaggregated impact assessments. • The buildup to the Dialogue, and elements within it, could include dedicated youth and child-rights consultations feeding into a child-safe AI design track. • Funding mechanisms to enable participation in the conference could include victims and frontline organizations addressing TFGBV.

To foster meaningful, dynamic engagement, the Dialogue should go beyond traditional plenary statements: Workshop-based deliberation. The July 2025 AI for Development workshop in Thimphu offers a tested model: bringing together government officials, technical experts, and end-users to co-design AI applications for concrete national priorities such as tourism platforms, agricultural diagnostics from satellite imagery, and entrepreneurship matching. Embedding similar workshop formats within the Dialogue—organized around real-world use cases rather than abstract principles—would generate practical, replicable insights. Live demonstrations and sandboxes. Showcasing AI tools in action—such as LLM chatbots, summarizers, matching applications, and biodiversity-monitoring systems—could allow participants to assess capabilities, limitations, and governance implications first-hand. Scenario exercises on AI risk. Tabletop simulations involving states and AI companies on hypothetical AI safety incidents can build the muscle memory needed for the proposed mechanism through which companies share safety testing results before model release, and through which member states coordinate on AI-related incidents. Tracks for identifying public commitments. Voluntary initiatives—such as a global set of priorities on gender-responsive AI auditing or child wellbeing—could be developed with a set of proposed metrics, to be further developed by the Independent International Scientific Panel on AI. Regional satellite sessions. Hosting parallel or preceding regional sessions, particularly in the Global South, would broaden participation beyond New York and Geneva and surface infrastructure realities—gaps in fiber, compute, energy, and affordability—that shape AI access. Cross-stakeholder co-drafting. An initiative towards commonly agreed standards on child-safe AI design or addressing TFGBV could be drafted in mixed groups of regulators, AI developers, civil society, and affected communities, as a starting point towards Member State and UN action. These formats prioritize doing over declaring, and would help the Dialogue produce cumulative, tangible outcomes.

The UAE highlights several examples that offer concrete contributions to effective AI governance: The UAE's Federal Decree-Law No. 26 of 2025 on Child Digital Safety. This framework provides a potential model for international consensus on minimum safety standards for AI systems targeting or accessible to minors, addressing exposure to harmful content, profiling, and exploitation. The UAE's Child Digital Safety Council. This Council can serve as the model for a proposed international knowledge-sharing platform on child digital safety and AI, enabling governments to exchange regulatory approaches, enforcement experiences, and technical standards for protecting children from AI-related harms. The Bhutan-UAE AI for Development workshop (Thimphu, July 2025). Co-convened by Bhutan and the UAE, this workshop demonstrated how AI applications-LLM chatbots, summarizers, matching tools, and satellite-imagery analytics-can advance national priorities in tourism, agriculture, and entrepreneurship while respecting local values such as Gross National Happiness. It is a replicable model of a South-South technology partnership that lower barriers to AI investment in smaller developing countries. Gender-responsive AI auditing. A voluntary initiative through which states and companies commit to addressing algorithmic impact assessments disaggregated by gender, in high-stakes sectors such as employment, education, healthcare, and public benefits, could seek translate international human rights commitments into measurable practice. AI for environmental management. AI tools could be used responsibly in environmental permitting, biodiversity assessments, deforestation tracking, and ecosystem mapping, and could serve as an example of how governance can ensure that AI supports expert assessments rather than replacing them-a principle that should be codified internationally. In addition, there could be benefit in international guidelines on sustainable AI infrastructure, including measures on fulfilling energy requirements through renewable energy and water recycling for cooling systems - practices being actively developed in the UAE.