Youth Policy Sandbox

The first Global Dialogue on AI Governance would be successful if it were to accomplish three major objectives. First, it will create an understanding of principles and priorities to guide international cooperation with regards to AI, anchored on human rights, international law, and sustainable development. Although the dialogue will not serve as a negotiation platform, reaching agreement on shared goals like avoiding harm, bridging inequalities, and ensuring human control would offer much-needed direction amidst the current disjointed efforts to govern AI. Second, success will be judged based on how inclusive participation can be. Inclusiveness here means not only inviting countries from the Global South, the youth, civil society, the technical community, and other affected parties as mere spectators to the dialogue, but also ensuring their active involvement. Third, there needs to be practical output from the Dialogue in the form of a report by the Chair on convergence points identified, gaps that need to be addressed, and possibilities for coordination with ongoing efforts within the United Nations system and regionally. In summary, the success of the Dialogue is tied to the capacity it creates to foster trust, avoid fragmentation, and allow for discussions on different governance mechanisms in the United Nations space.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

These priorities are indicative of the areas where immediate action and cooperation are required. Building capacity around AI is essential in order for the governance process to be relevant, especially for developing nations. Failure to develop capacity means that any global discussion around AI will simply exacerbate existing imbalances and will effectively exclude much of the world from influencing AI development. The respect and protection of human rights needs to be at the heart of AI governance. AI technologies continue to influence human rights related to privacy, equality, accessibility, and participatory democracy. Human rights law presents a globally accepted normative framework which can facilitate cooperation between states in addressing these issues. Transparency, accountability, and human oversight are crucial for building trust in AI technologies and enabling solutions when things go wrong. This would facilitate the implementation of human rights and would make the implementation of the above-mentioned values feasible and comprehensible to people affected by those measures. Lastly, interoperability is required in order to prevent division on a global scale. In view of growing numbers of different AI frameworks being developed within national contexts, discussions should be held in order to determine common grounds and areas for cooperation in the interest of smaller states. These priorities constitute the basis of an approach to AI governance on a global level.

The first cross-cutting issue relates to global equity regarding AI value chains, which include issues like data access, computing capacity, and power dynamics in the markets. Discussions of AI governance tend to emphasize the impacts in their later phases without giving much consideration to issues that determine who reaps the benefits of and bears the burdens of using AI technologies. A second cross-cutting issue relates to the environmental and climate impacts of AI, specifically, the energy, water, and material costs of developing and using large AI systems. The climate impacts of AI relate directly to questions of sustainable development and will disproportionately affect places that already suffer climate stress. There is a third cross-cutting issue related to AI governance in an international or transnational context, especially where it concerns migration management, humanitarian action, and border surveillance operations. This area poses jurisdictional and responsibility challenges that are beyond national mechanisms. Considering these cross-cutting issues in the AI governance dialogue can help increase its relevance.

Gaps in governance within the identified thematic areas are already contributing to challenges and opportunities for the Philippines and more broadly Southeast Asia. The foremost challenge stems from disparities in the ability to develop and regulate AI technology. Despite increasing AI deployment in industries such as banking, government service delivery, content moderation, and surveillance technologies, there is often limited knowledge and regulatory infrastructure to effectively assess risks, ensure protections, or participate in international standards development processes. This disparity leads to an increased reliance on third-party AI products, which may not take into account the domestic laws, languages, and socio-economic context. Growing human rights risks exist in the use of AI-based technologies in social protection eligibility, law enforcement, content moderation, and migration administration. Privacy, non-discrimination, procedural fairness, and redress remain pressing issues in the development and implementation of these tools. Lack of accountability measures makes it difficult for individuals to raise grievances when they experience harms resulting from AI use. Human rights norms apply, yet their implementation is patchy. Meanwhile, fragmented forms of global and regional governance pose challenges to policymakers and regulatory bodies as well. Standards diverging among various jurisdictions can pose a problem for international collaboration and lead to compliance issues for both national governments and small businesses. In the context of Southeast Asian countries, lack of participation in setting these standards poses a risk of promoting poor norms that do not fit their needs for development. Nevertheless, there are significant opportunities here as well. Potential advantages of artificial intelligence in terms of improving the efficiency of public sector, minimizing disaster risks, facilitating access to language, and economic empowerment could emerge provided adequate governance and inclusivity are considered. Regional emphasis on digital cooperation and the interest in AI governance by ASEAN could help leverage these opportunities. By closing the gaps in governance through international cooperation and capacity building, AI could be leveraged successfully and its threats addressed properly.

The AI Dialogue can play a critical role as a bridging and coordinating platform within the international AI governance landscape. At a time when multiple national, regional, and sector‑specific initiatives are emerging, the Dialogue can help reduce fragmentation by providing a space for governments and stakeholders to exchange perspectives, identify areas of convergence, and clarify points of divergence in a transparent and constructive manner. Importantly, the AI Dialogue can function as a trust‑building forum. By convening states at different stages of AI development alongside non‑state actors, the Dialogue can foster mutual understanding between technology‑producing and technology‑adopting countries. This is particularly valuable for addressing concerns related to capacity gaps, equitable access to AI benefits, and asymmetries in agenda‑setting power. The Dialogue can also advance cooperation by connecting principles to implementation. Through experience‑sharing on domestic policies, regulatory approaches, and governance tools, participants can learn how high‑level commitments—such as human rights protection, accountability, and human oversight—are translated into practice across diverse contexts. This can inform adaptive and context‑sensitive governance models rather than one‑size‑fits‑all solutions. In addition, the AI Dialogue can serve as a gateway for inclusive participation in global AI governance. By elevating perspectives from developing countries, youth, civil society, and affected communities, the Dialogue can help ensure that international cooperation reflects a broader range of social, cultural, and developmental realities. Ultimately, the value of the AI Dialogue lies not in replacing existing initiatives, but in strengthening coherence, inclusivity, and shared purpose across global AI governance efforts.

The AI Dialogue should build upon a diverse ecosystem of existing initiatives, while providing added value through coordination, inclusivity, and political visibility. Existing efforts include UN processes and soft‑law instruments addressing digital technologies, such as human rights‑based guidance on AI, UN agency initiatives on ethical and responsible AI, and capacity‑building programs supporting digital governance in developing countries. Regional frameworks—such as emerging AI governance discussions within ASEAN, the African Union, and the European Union—also offer valuable insights into context‑specific approaches. Multi‑stakeholder platforms involving governments, the private sector, civil society, and the technical community have generated practical tools, best practices, and technical standards. However, these initiatives often operate in parallel, with limited opportunities for cross‑regional learning or input from less‑resourced actors. The AI Dialogue can help connect these efforts, creating visibility for existing work while identifying gaps and overlaps. The added value of the AI Dialogue lies in its universal convening power under the United Nations. Unlike smaller or region‑specific forums, it can bring together a politically representative set of actors while maintaining openness to non‑state participation. This creates opportunities to align technical, normative, and policy discussions in a single space. Additionally, the Dialogue can contribute by synthesizing lessons learned across initiatives and articulating shared priorities for future cooperation, without duplicating mandates. By amplifying underrepresented perspectives and fostering links between technical practice and international policy, the AI Dialogue can enhance coherence and legitimacy across the global AI governance landscape.

Various stakeholders have different roles to play in the AI Dialogue in terms of their unique expertise, lived experiences, and institutional perspectives. The government is vital in establishing regulatory priorities, aligning domestic policy approaches, and defining areas that demand international collaboration. Civil society groups can participate by focusing on human rights considerations, ethical aspects, and the perspectives of disadvantaged individuals who suffer the adverse consequences of AI technologies. Non-state participants such as the business community and experts from the technical community can provide relevant insights regarding AI design, implementation, risk assessment, and feasibility analysis of regulatory interventions. Academics and research centers can provide empirical data and interdisciplinary perspectives on social consequences. In order to maximize the contributions of various stakeholders, the Dialogue must adopt a multi-tiered structure of both plenary and breakout sessions that will focus on specific regulatory issues. Multi-stakeholder panels should not cluster participants based on common interests but rather balance their representation. Separate speaking slots should be allocated to non-state actors like young people and delegates from the Global South to minimize tokenization. The Dialogue could also gain from preparatory papers prior to the session, which would help in holding discussions based on sound knowledge and reduce duplication of effort during sessions. Unofficial roundtable discussions or small groups chaired by neutral mediators could promote constructive discussions and problem-solving among stakeholders. Lastly, it is imperative that the Dialogue's process provide opportunities for follow-up. These may take the form of documenting convergences, best practices, and areas where further action is needed. Through this process, stakeholders' participation can extend beyond the dialogue, thus enhancing its legitimacy.

Global discussions on AI governance continue to underrepresent several critical voices and perspectives. These include stakeholders from developing countries, particularly small states and least developed countries; grassroots civil society organizations; youth; indigenous peoples; linguistic minorities; and communities disproportionately affected by AI‑enabled decision‑making, such as migrants, informal workers, and persons with disabilities. Underrepresentation often stems from structural barriers, including limited financial resources, technical capacity gaps, language constraints, and lack of access to international policy spaces. As a result, governance frameworks risk reflecting the priorities and assumptions of a narrow group of technology‑producing states and actors, rather than the realities of diverse social and economic contexts. To address this gap, inclusion must go beyond open invitations. Targeted support measures—such as travel and participation funding, interpretation beyond the UN's working languages, and capacity‑building prior to the Dialogue—are essential to enable effective participation. Simplified and accessible consultation formats, including online and regional preparatory dialogues, can further broaden engagement. Thematic sessions should intentionally integrate perspectives from affected communities, not only as "impact stories" but as contributors to policy design. Youth and community representatives should be included as speakers and discussants, rather than limited to parallel forums. By embedding inclusion into both the design and resourcing of the AI Dialogue, the process can better reflect global diversity and ensure that AI governance responds to those most affected by its deployment.

Innovative engagement formats are essential for moving beyond formal statements toward genuine dialogue and collective problem‑solving in AI governance. One promising approach is scenario‑based discussions, where participants jointly examine realistic AI governance challenges—such as cross‑border data use or automated decision‑making in public services—and explore policy responses from different stakeholder perspectives. Facilitated roundtables with mixed stakeholder representation can encourage candid exchange by placing governments, civil society, technical experts, and private actors in the same small‑group setting. Neutral facilitators and clearly defined discussion questions help balance power dynamics and ensure all voices are heard. The Dialogue could also incorporate policy clinics or solution labs, where participants present emerging practices or regulatory approaches and receive peer feedback. This format promotes mutual learning and highlights practical governance tools rather than abstract principles. Digital participation tools—such as real‑time polling, moderated online Q&A, and collaborative digital whiteboards—can enhance inclusivity, particularly for participants joining remotely or from different time zones. These tools allow broader participation without sacrificing interactivity. Finally, intergenerational dialogues pairing youth leaders with senior policymakers could introduce fresh perspectives while fostering mutual understanding. When combined with clear documentation of insights and outcomes, these innovative formats can make the AI Dialogue more dynamic, inclusive, and action‑oriented, strengthening its role as a global platform for responsible AI governance.

Several potential policies, processes, and technologies already exist that may contribute to improved AI governance. From an international perspective, the development of guidance on the use of AI grounded in human rights considerations-such as AI impact assessments grounded in international human rights law-offer tools for identifying, preventing, and mitigating harm before AI systems are implemented. Through implementation in public procurement and regulatory processes, this process can turn lofty ideals into binding requirements. In addition, regulatory sandboxes and governance pilots are promising policy solutions. Through regulatory sandboxes, governments, innovators, and regulators are able to experiment with the use of AI technology under controlled conditions, enabling both learning by doing as well as necessary safeguards and oversight. This solution is especially relevant for countries without significant capacity to regulate AI. Multi-stakeholder platforms involving governments, civil society organizations, technical experts, academia, and the private sector have shown their worth for sharing knowledge and norms. Open consultations and participatory policy-making enable the integration of different social, cultural, and linguistic contexts into AI governance, including those from the Global South. From the perspective of institutional practice, public-sector AI registries and transparency portals where governments provide information about the AI systems used in public services foster accountability and build trust. Alongside mechanisms for complaints and remedies, these tools facilitate human supervision and remedy access. Lastly, regional cooperation structures, such as initiatives aimed at promoting interoperability, learning from each other, and developing capacities jointly, can be scaled up and used as instruments against fragmentation in governance. Regional guidelines that take into account global principles but consider local specifics will avoid duplication of efforts and minimize compliance costs, giving small countries more leverage in global conversations on AI governance. All these examples show how AI governance can be successfully implemented through the combination of normative commitments and context-specific tools.