The United Nations is building a more inclusive framework for AI governance, but countries with limited technological infrastructure still face unequal influence over the systems transforming their economies.
Forty experts now serve on the United Nations’ Independent International Scientific Panel on Artificial Intelligence, selected from more than 2,600 candidates to assess the technology’s risks, capabilities and social consequences. The panel was established in 2025, with members appointed in February 2026, and presented its preliminary findings at the first Global Dialogue on AI Governance in Geneva in July. The effort reflects a growing recognition that artificial intelligence cannot be governed exclusively by the countries and corporations developing its most powerful systems. For nations across Africa, Latin America, Asia and the Caribbean, the debate is not limited to safety. It also concerns economic opportunity, technological dependence and whether they will help establish the standards that increasingly shape their societies.
Participation Is Expanding. Influence Remains Uneven.
The first UN Global Dialogue brought together representatives from 163 countries, along with more than 3,000 participants and over 1,000 stakeholder organizations. Its scientific panel includes experts from different geographic regions and levels of technological development, with a mandate to provide evidence that all member states can use. That structure addresses a persistent imbalance in international technology policymaking: countries with fewer research institutions, computing resources and specialized personnel often have less capacity to evaluate systems developed elsewhere. A shared scientific foundation can improve their participation, although access to information does not automatically translate into equal negotiating power.
The concentration of AI development among a relatively small number of technology companies creates additional challenges. Many governments depend on foreign cloud providers, imported computing equipment and commercial models trained primarily for larger markets. Those dependencies can affect costs, data protection, language accessibility and the ability to adapt systems to local conditions. A government deploying AI in schools or healthcare may have limited influence over the underlying model’s design, updates or commercial terms. For developing economies, participation in international rulemaking is therefore connected to the practical question of how much control they retain over public services and digital infrastructure.
The Economic Stakes Extend Beyond Technology
AI could improve agricultural planning, medical diagnostics, education and administrative efficiency in countries where resources are constrained. But realizing those benefits requires reliable electricity, connectivity, technical expertise and institutions capable of evaluating the technology. Countries without those foundations may become primarily customers rather than producers of advanced systems. Their businesses could face higher costs or limited access to products optimized for wealthier markets, while local workers encounter automation without corresponding investment in retraining. The economic opportunity is substantial, but it is not distributed automatically.
Language and culture also belong in the discussion. AI systems trained on disproportionately English-language or Western data may perform less reliably in underrepresented languages and local contexts. This can influence educational materials, translation, public information and commercial services. Governments need the capacity to test those systems before adopting them at scale. The UN’s scientific framework could help establish common evaluation methods, but national institutions will still need funding and expertise to apply them.
The UN Is Building an Evidence-Based Framework
The UN panel’s preliminary report examines AI across seven broad areas, including economic implications, human rights, security, environmental consequences and social applications. Its role is scientific rather than legislative: it produces assessments intended to inform governments, not binding regulations. That limitation is significant. International dialogue can identify shared principles and improve cooperation, but implementation remains dependent on national governments and other institutions. Disagreements over innovation, trade, security and corporate accountability are unlikely to disappear simply because countries share access to research.
The UN has also recognized that governments may need to act before every technological risk is fully understood. Waiting for conclusive evidence can leave institutions responding only after harm has occurred. Acting too aggressively, however, can restrict useful innovation or create regulatory burdens that smaller economies struggle to meet. Developing countries therefore have an interest in rules that are proportionate, adaptable and supported by technical assistance. International governance must account for differences in institutional capacity rather than assuming every country can enforce identical requirements with equal resources.
What Equal Participation Would Require
A meaningful role for developing countries involves more than representation at conferences. It requires investment in local research, affordable computing infrastructure, access to technical expertise and mechanisms for evaluating commercial AI systems independently. Governments also need negotiating capacity when procuring technology for public services. Regional cooperation could help smaller countries pool expertise, establish common standards and avoid becoming individually dependent on a limited number of suppliers.
The second Global Dialogue is scheduled for May 2027 in New York, giving governments another opportunity to assess progress. Between now and then, the measure of international cooperation will be whether participation produces practical capabilities and greater institutional independence. The world is moving toward a more inclusive conversation about artificial intelligence. Whether that conversation changes who controls the technology remains unresolved.