Advanced Science and Technology Institute

A successful first Global Dialogue on AI Governance should produce more than broad statements of goodwill. It should establish a practical foundation for sustained cooperation, especially between countries with very different levels of AI capability, infrastructure, and regulatory maturity. First, it should create a shared understanding of priority risks and opportunities: safety, misinformation, privacy, labor impacts, cybersecurity, bias, concentration of power, and the responsible use of AI in public services. Success would mean agreeing that AI governance is not only about preventing catastrophic risks, but also about ensuring that ordinary citizens, developing economies, and marginalized communities benefit from AI. Second, the Dialogue should identify concrete areas for international coordination. These may include interoperable standards, model evaluation and auditing practices, incident reporting mechanisms, data governance principles, compute and infrastructure access, and capacity-building support for developing countries. A useful outcome would be a roadmap that countries can adapt locally without forcing a one-size-fits-all framework. Third, it should elevate inclusivity. Global AI governance cannot be shaped only by technologically advanced economies or large private companies. The voices of the Global South, academia, civil society, startups, public-sector users, and affected communities should be meaningfully represented. Finally, the Dialogue would be successful if it leads to continuing mechanisms: working groups, shared research agendas, technical assistance programs, and regular follow-up meetings with measurable commitments. In short, success means moving from principles to implementation: building trust, aligning standards where possible, respecting national contexts, and ensuring that AI becomes a tool for human development rather than another source of inequality.

• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Interoperability of governance approaches
• Social, economic, ethical, cultural, linguistic and technical implications of AI

I selected these four areas because they represent the minimum foundation for responsible and inclusive AI governance. First, safe, secure, and trustworthy AI is urgent because AI systems are increasingly being deployed in critical sectors, including government, education, health, finance, disaster risk reduction, and public services. Without safety, security, and trust, adoption will either become reckless or resisted by the public. Second, the social, economic, ethical, cultural, linguistic, and technical implications of AI must be prioritized because AI governance should not be limited to technical performance. For countries like the Philippines and other developing economies, AI must account for local languages, cultural contexts, labor impacts, digital divides, and unequal access to data, infrastructure, and skills. Third, interoperability of governance approaches is essential. Countries will develop their own AI policies, but fragmented rules can create confusion, duplication, and barriers to cooperation. Shared principles, compatible standards, and mutual learning can help countries govern AI while still respecting national contexts. Fourth, transparency, accountability, and human oversight are necessary to ensure that AI systems remain explainable, auditable, and subject to responsible decision-making. This is especially important when AI affects rights, access to services, public resources, or institutional decisions. Together, these priorities support a balanced approach: enabling innovation while managing risks, protecting people, strengthening institutions, and ensuring that AI contributes to inclusive and sustainable development rather than deepening existing inequalities.

Yes. One important cross-cutting issue is equitable access to AI infrastructure, especially compute, quality datasets, cloud services, talent, and evaluation tools. Many governance discussions focus on rules and principles, but countries with limited infrastructure may remain rule-takers rather than meaningful participants in AI development. Without addressing the compute and data divide, global AI governance may unintentionally deepen technological dependency. A second issue is AI sovereignty and local relevance. Countries need the capacity to develop, adapt, and evaluate AI systems that reflect their own languages, cultures, legal systems, public-sector needs, and development priorities. This is especially important for multilingual societies, indigenous communities, and countries whose realities are underrepresented in global datasets. Third, there is a need to address AI market concentration and dependency on a few dominant platforms. Governance should consider how to prevent excessive concentration of power over models, infrastructure, standards, and data. This affects competition, national resilience, public procurement, and the ability of smaller economies, startups, universities, and public institutions to innovate. Fourth, environmental sustainability should be treated as a cross-cutting concern. Large-scale AI development requires significant energy, water, and hardware resources. Responsible AI governance should include attention to green computing, efficient model development, lifecycle assessment, and the environmental costs of AI deployment. Finally, there should be stronger attention to implementation capacity. Many countries can endorse ethical principles, but lack institutions, technical expertise, testing facilities, and enforcement mechanisms. Global AI governance should therefore support practical toolkits, capacity-building, shared evaluation resources, and mechanisms for translating principles into real institutional practice.

Governance gaps in the selected thematic areas affect the Philippines, ASEAN, and the public-sector research community in very practical ways. The most significant challenge is uneven readiness. AI adoption is accelerating, but institutions often lack clear policies, technical standards, evaluation methods, procurement guidance, and skilled personnel to assess whether systems are safe, secure, trustworthy, transparent, and accountable. This can lead either to risky deployment of AI tools without sufficient safeguards, or to excessive caution that slows innovation. A second challenge is the digital and capability divide. Many agencies, local governments, schools, MSMEs, and research institutions still face constraints in data quality, compute resources, cybersecurity maturity, and AI talent. These gaps make it difficult to participate meaningfully in AI development and governance, especially when global standards are shaped mainly by technologically advanced economies and large private companies. There are also important social and cultural challenges. AI systems may not adequately reflect Philippine languages, local contexts, public-sector workflows, or the realities of vulnerable communities. This raises risks of exclusion, bias, misinformation, and poor service delivery. At the same time, these developments create major opportunities. Stronger AI governance can build public trust, improve responsible adoption in government, and support better services in health, education, disaster risk reduction, agriculture, transportation, and public administration. Interoperable governance approaches can help the Philippines align with regional and global standards while still protecting national priorities. There is also an opportunity for the Philippines and ASEAN to help shape a more inclusive AI governance agenda—one that emphasizes capacity-building, local language technologies, public-interest AI, open collaboration, and practical implementation. The goal should be to move from principles to usable frameworks, tools, and institutions that allow countries not only to consume AI, but to govern, adapt, and develop it responsibly.

The AI Dialogue can play a crucial role as a sustained platform for building trust, coordination, and practical cooperation on AI governance. Its value should not only be in producing declarations, but in helping countries move from broad principles to implementable actions. First, the Dialogue can create a shared understanding of AI risks, benefits, and governance priorities across different national contexts. This is important because countries vary widely in their level of AI readiness, infrastructure, technical capacity, regulatory maturity, and exposure to risks. Second, it can promote interoperability among governance approaches. Countries do not need identical laws, but they need compatible principles, standards, evaluation methods, and risk management practices so that AI systems can be developed and used responsibly across borders. Third, the Dialogue can support capacity-building for developing countries. International cooperation should include technical assistance, shared testing and evaluation tools, access to expertise, support for local language technologies, and mechanisms to reduce the AI infrastructure and data divide. Fourth, it can serve as a venue for inclusive agenda-setting. The voices of developing economies, academia, civil society, public-sector institutions, startups, and affected communities should help shape global AI governance, not only large technology firms and advanced economies. Finally, the Dialogue can establish continuing mechanisms such as working groups, common research agendas, incident-sharing channels, policy sandboxes, and regular follow-up reviews. This would help ensure that cooperation remains practical, adaptive, and accountable as AI technologies rapidly evolve. In short, the AI Dialogue can help align global efforts while respecting national contexts, enabling countries to govern AI in ways that are safe, trustworthy, inclusive, and beneficial for sustainable development.

The AI Dialogue should build upon existing efforts rather than create another isolated process. These include the UN Global Digital Compact and the UN High-level Advisory Body on AI, which provide a multilateral foundation for addressing AI risks and sharing benefits globally. It should also connect with the OECD AI Principles and GPAI, UNESCO's Recommendation on the Ethics of AI, the G7 Hiroshima AI Process, the International Network of AI Safety Institutes, and regional mechanisms such as the ASEAN Guide on AI Governance and Ethics. These initiatives already provide useful principles, standards, technical work, and regional experience. The added value of the AI Dialogue is its universality and convening power. It can bring together Member States, developing countries, regional organizations, academia, civil society, technical experts, and the private sector in a more inclusive setting. It can help identify overlaps, reduce fragmentation, and translate high-level principles into practical tools, capacity-building programs, shared evaluation methods, incident-sharing mechanisms, and interoperable governance approaches. For countries like the Philippines and others in ASEAN, the Dialogue can amplify priorities that are sometimes underrepresented in global AI debates: local language technologies, public-sector adoption, compute and data access, AI skills development, MSME participation, disaster resilience, and safeguards against widening inequality. In short, the AI Dialogue can serve as a bridge: connecting global, regional, and technical initiatives; ensuring that developing countries are not left behind; and moving international AI governance from scattered principles toward coordinated, practical, and inclusive implementation.

Different stakeholders can contribute by bringing both policy perspectives and practical experience. Governments can identify national priorities, regulatory gaps, public-sector use cases, and areas where international cooperation is needed. International and regional organizations can help align the Dialogue with existing frameworks and avoid duplication. Academia and research institutions can provide evidence, independent evaluation, technical expertise, and foresight on emerging risks and opportunities. The private sector can share implementation experience, safety practices, standards work, and innovation pathways. Civil society can represent affected communities and raise concerns on rights, inclusion, accountability, labor, and social impacts. Developing countries, local governments, MSMEs, and community organizations should also be given meaningful space, since they often experience AI's impacts without having equal influence over its design and governance. For the format, the AI Dialogue should combine high-level policy discussions, technical working groups, regional consultations, multi-stakeholder roundtables, and practical policy clinics focused on areas such as safety, transparency, accountability, interoperability, data governance, capacity-building, and equitable access to AI infrastructure. It should also include regular follow-up mechanisms with measurable outputs, timelines, and reporting, so that the Dialogue becomes a continuing process rather than a one-time event. Its success will depend on whether it can produce practical guidance, connect countries to technical support, reduce fragmentation, and ensure that AI governance is shaped not only by advanced AI developers, but also by those most affected by AI deployment.

Several voices remain underrepresented in global AI governance discussions. First are developing countries and small economies, especially those from the Global South. Many are affected by AI systems and global standards, but have limited influence over how these systems and rules are designed. Their priorities—capacity-building, affordable compute, data access, digital public infrastructure, local innovation, and avoiding technological dependency—should be more central. Second are local communities, indigenous peoples, linguistic minorities, persons with disabilities, workers, teachers, students, farmers, MSMEs, and public-sector users. These groups experience AI's real-world effects in employment, education, public services, culture, and access to opportunities, but are often absent from high-level policy forums. Third are researchers, startups, and civil society organizations from countries with limited AI resources. Their perspectives are important because AI governance should not be shaped only by large technology firms, major economies, and well-funded institutions. They can be included by designing the AI Dialogue as a genuinely multi-stakeholder and regionally balanced process. This means providing funded participation for developing-country representatives, holding regional and local consultations before global meetings, creating dedicated tracks for the Global South, youth, civil society, academia, MSMEs, and affected communities, and allowing inputs in multiple languages. The Dialogue should also support practical participation, not just symbolic representation. Underrepresented stakeholders should be involved in agenda-setting, drafting recommendations, working groups, case studies, policy clinics, and follow-up mechanisms. Inclusion should be measured by whether their concerns shape actual outputs—such as capacity-building programs, local language AI initiatives, public-interest AI safeguards, and equitable access to infrastructure—not merely by whether they were invited to attend.

The AI Dialogue should move beyond traditional speeches and panel discussions. To foster meaningful engagement, it should use formats that allow stakeholders to work together on concrete problems. One effective format would be policy clinics, where countries present real governance challenges—such as AI procurement, model evaluation, data governance, or public-sector deployment—and receive practical feedback from technical experts, regulators, civil society, and peers. Another useful format would be scenario-based exercises or simulations. Participants could examine situations such as AI-enabled misinformation during elections, algorithmic bias in public services, cross-border AI incidents, or cybersecurity risks. This would help translate abstract principles into practical decision-making. The Dialogue could also include regional breakout sessions to surface local priorities, especially from ASEAN, the Global South, small island states, and developing economies. These sessions should feed directly into the main outcomes, not function as side conversations. Multi-stakeholder roundtables with small, balanced groups would also be valuable. These should include governments, academia, industry, civil society, youth, workers, MSMEs, and affected communities, with facilitators ensuring that less powerful voices are heard. The Dialogue could further use innovation showcases and public-interest AI demonstrations, highlighting responsible AI applications in health, education, disaster resilience, agriculture, local language technologies, and public services. Finally, an online participation platform could collect written inputs, rank priorities, share draft recommendations, and allow stakeholders who cannot travel to contribute meaningfully. The most important design principle is that engagement should produce outputs: practical toolkits, recommendations, case studies, working group agendas, and follow-up actions. The Dialogue should not only ask people to speak; it should enable them to shape decisions.

Examples of effective AI governance include risk-based regulation, management standards, public-sector guidance, and practical testing platforms. The EU AI Act is a useful example of a risk-based approach: stricter obligations are applied to higher-risk AI systems, while unacceptable uses are restricted or prohibited. This helps match regulation to actual levels of harm rather than treating all AI systems the same. The NIST AI Risk Management Framework offers a practical model for organizations to identify, assess, measure, and manage AI risks across the AI lifecycle. It is useful because it can be adapted by governments, companies, and research institutions without being overly prescriptive. ISO/IEC 42001 is another important example because it turns AI governance into an organizational management system. It encourages institutions to establish processes for accountability, transparency, risk management, monitoring, and continual improvement. At the regional level, the ASEAN Guide on AI Governance and Ethics provides a practical reference for organizations designing, developing, and deploying AI, while supporting alignment and interoperability across ASEAN. Other useful approaches include AI impact assessments, algorithmic audits, public procurement guidelines, incident reporting systems, regulatory sandboxes, data governance frameworks, and human oversight requirements. For developing countries, concrete solutions should also include shared testing facilities, open-source evaluation tools, local language datasets, capacity-building programs, and regional AI centers of excellence. The best governance approaches are not merely legal documents. They combine principles, standards, technical tools, institutional capacity, and accountability mechanisms so that AI can be deployed safely, transparently, and inclusively.