Southern Voice

An equitable future of AI is an important outcome of improved international development cooperation. The first Global Dialogue on AI Governance can advance, in this respect, global multi-stakeholder commitment towards some of the broader areas of priority for AI to be a lever of accelerated sustainable development progress. This can include: Promoting funding and investment to the Global South, to enable the development of AI tools responding to local challenges; Emphasising the development of AI and data governance capacities Surfacing successful international development cooperation models that can be built upon to ensure inclusive, equitable governance where applicable, and develop fair supply chains. In addition, the following outcomes, based on recommendations from experts within the Southern Voice network, respond to some of the areas of priority outlined in our responses to the next questions in this survey. For many countries of the Global South, developing appropriate regulatory, and monitoring and detection mechanisms for AI misinformation is complex: fast-evolving context, overworked bureaucracies, limited capacity. A collaborative approach to designing solutions to these issues would be beneficial. Towards this, the Global Dialogue could establish a collaborative track, with adequate funding, for the design of solutions to safeguard information integrity across Global South contexts. In addition, there is an urgent need to improve monitoring and accountability mechanisms aimed at safeguarding information integrity. Towards this, the Global Dialogue could define modalities for a multi-stakeholder process to establish a viable accountability mechanism for AI platforms. Finally, the Global Dialogue could ensure that its outcomes enable an integrated approach to AI governance – which for example takes into consideration the implications of AI on energy and water consumption, as well as on climate change more broadly.

• Safe, secure and trustworthy AI
• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches

Safe, secure and trustworthy AI: With rapid changes in technology and limited capacity, governments in many Global South contexts are struggling to design appropriate regulations and guidelines for AI systems. This has repercussions for 1) information integrity, with misinformation a growing threat in the Global South; and 2) online safety, with technology-facilitated abuse being a priority area of attention that requires improved governance frameworks focused on rights-based regulation on privacy. Social, economic, ethical, cultural, linguistic and technical implications of AI: AI models mimic human biases, as they are created from existing data that may reflect prejudices and inequalities. Data needs to reflect society's diversity, including languages and gender realities. Further, AI and automation is driving unemployment, primarily among young people, and rapidly increasing climate adverse effects. In addition, AI is intensifying the demand for critical minerals, which is reinforcing the interdependence of digital and extractive value chains. AI capacity building: Successful AI deployment requires long-term investments in capacity building at various levels necessary to enable safe, inclusive, equitable adoption. This includes specialised training for engineers, data scientists, and regulators. Further, AI's effective use in decision-making requires updating the skills of existing public servants - to fulfill necessary functions such as data specialists, prompt engineers, machine learning experts. Finally, educational institutions can strengthen human capital by preparing young people with skills related to AI and data analytics. Interoperability of governance approaches: As highlighted in a recent statement by BRICS leaders, principles of equitable and inclusive AI global governance "should be applied through the development of [interoperable, transparent, and consensus-based] standards and protocols." Further, data systems should be built in a way that allows for responsible sharing across institutions, through common standards and cleared stewardship. There is a need to invest in local data infrastructures through registries, interoperable systems, and quality control.

AI & Climate Change: AI is both an opportunity to enhance climate action, and a threat. A cohesive climate action strategy can use AI in favor of sustainable development, but it is also important to recognise the risk and negative impacts of AI-driven technologies on climate change. AI's huge infrastructure needs are reflected in data centers, which require significant amounts of energy and water to run, and are fundamental to AI progress. However, AI also offers increased opportunities for environmental data analysis, geospatial monitoring, and disaster risk forecasting. There is potential for AI to be embedded in national and business strategies for climate change, in such a way that regulatory sanctions and economic incentives are used to promote a more environmentally-friendly deployment of AI. Policies should prioritise climate preventive strategies and encourage climate-focused digital entrepreneurship that can develop AI-driven solutions for environmental challenges. Sovereignty, geopolitics, & supply chains: There is a risk of Global South countries being locked in extractivist resource-supplier relationships, with strategic vulnerabilities that can be exploited. AI is recognized as a driver for energy transition, but its implementation is amplifying structural inequalities in production chains and consolidating new dependencies (e.g. specialized equipment and supply chains are concentrated in certain regions, export controls on advanced GPUs, and externalised data governance through hyperscale operators that determine the jurisdictional rules, processing, and storage of data). Data readiness: AI systems need structured, reliable, local data to be effective. Countries of the Global South need to build data ecosystems that make AI work for everyone, including women and marginalised regions. Therefore, data readiness should be prioritised as much as AI readiness.

Innovation in public administration: The adoption of AI in public administration has enormous potential as it can help maximize the delivery of public services and prevent new forms of inequality. However, AI models can be biased if the data they use have prejudices and inequalities, which is why these tools need to be more transparent and accountable at the public level. AI as a lever for sustainable development: Countries and regions across the Global South need to operationalise AI as a lever for equitable, transformative progress towards sustainable inclusive development – including through strengthened economic growth. For this, it is important to think of AI beyond the extractivist logic, where cycles of technological and economic dependence are created in the Global South. Instead, AI can be a primary tool for innovation, tech, regional integration, and infrastructure investment. Inequalities: Governance frameworks should account for the distributional impacts of AI, such as the 'winners and losers' across economies and labour markets. Regional, age, gender, and socioeconomic disparities also exist in digital access and skills, which can deepen inequalities and limit access to AI. Regulatory mechanisms should be put in place early before harmful dynamics become entrenched and extremely difficult to reverse. Information integrity: generative AI is unregulated and freely available, which increases the ease to spread false information and, at the same time, the urgent need to improve monitoring mechanisms that can safeguard information integrity. To foster global information integrity, governments must promote media literacy, tech companies must be made accountable, fact-checking mechanisms must be expanded, and the Global South's information ecosystem must be strengthened.

Fragmentation is a key concern for the effectiveness of international development cooperation. The AI Dialogue can be a driver of coordinated action across countries, tech developers, civil society, and other actors in order to build synergies and greater alignment. An inclusive multi-stakeholder approach should be embedded through all steps leading up to the Dialogue, during the Dialogue, and in the follow-up to the Dialogue – bringing together policymakers, research ecosystems including social and environmental scientists, civil society, responsible private sector actors, and multilateral partners. Inclusive governance is also a priority for strengthened international development cooperation. The model of the Dialogue is well-positioned in this regard – though there would be value in defining viable pathways to ensure that other spaces in which AI is being discussed strengthen their approaches to inclusive governance, where relevant stakeholders can shift from being consulted, to being an active part of the definition of guiding principles and priorities for international cooperation on AI governance. It is also essential to prioritise the concerns that limit the benefits that Global South countries can derive from the use and development of AI. These include computing capacity, energy grids, and connectivity. The Dialogue's approach, which actively emphasises capacity questions, will help ensure that investment in infrastructure, for example, remains a core question. The format of the Dialogue may enable discussions to effectively surface areas of mutual benefit, and partnership models that can support strengthening these investments. Current mechanisms for digital governance reflect geopolitical dynamics and competing visions of the global digital order: the AI Dialogue can pave the way for the creation of regulatory models that strengthen privacy and data protection, without sacrificing innovation or security. Finally, the Dialogue can help advance South-South Cooperation efforts for knowledge sharing on AI governance issues and solutions.

The AI Dialogue should actively build on existing initiatives that have demonstrated what inclusive, locally-grounded AI governance can look like in practice. Southern Voice's network of 71 think tanks across 35 countries in Africa, Asia, Latin America, and the Caribbean represents an established infrastructure for surfacing Global South perspectives on AI governance. The Dialogue should treat such networks as active partners — not just consulted, but co-designing priorities and recommendations. At the national level, initiatives like Solidar Tunisie's Observatoire Tunisien des Politiques Publiques demonstrate how local data infrastructure can strengthen transparency and evidence-based policymaking — a model worth scaling and connecting to AI readiness efforts globally. LATAM-GPT illustrates how Global South actors can develop AI tools tailored to local languages and contexts. The Dialogue should amplify and mobilise funding for similar initiatives. Finally, BRICS governance principles (emphasising interoperability, transparency, and consensus-based standard-setting) offer a foundation for building regulatory frameworks that reflect a broader range of national interests.

Think tanks across the Global South can provide essential, locally-grounded evidence and perspectives, to ensure that the AI Dialogue promotes an inclusive, equitable future for AI that can accelerate the achievement of key sustainable development objectives. Southern Voice is a network of 71 think tanks across 35 countries of Africa, Asia, Latin America, and the Caribbean. Over the coming months, we will build on the insights shared here (and others), to shape concrete recommendations to strengthen the role of AI as a lever of transformation and sustainable development across the Global South, and mitigate its negative impacts. The insights shared in this survey are grounded in recent dialogues with experts in the Southern Voice network, as well as in recent publications by Southern Voice, and experts within the following network member institutions: Center for Study of Science, Technology and Policy (CSTEP), India; Centre for the Study of the Economics of Africa (CSEA), Nigeria; Centro de Implementación de Políticas Públicas para la Equidad y el Crecimiento (CIPPEC), Argentina; Institute for Global Dialogue (IGD), South Africa; Plataforma CIPÓ, Brazil; Policy Center of the New South (PCNS), Morocco; Public Affairs Centre (PAC), India; Solidar Tunisie, Tunisia; Sustainable Development Policy Institute (SDPI), Pakistan. We are strongly interested in engaging with the Dialogue – which can include sharing joint perspectives, orrecommending specific experts for targeted discussions. Broader civil society Our recent exchanges also stress the importance of civil society's role, not only as implementers, or as a constituency to be consulted, but as partners in co-creation and decision-making. This is an important consideration for the format of the dialogue, ensuring that sufficient room is given to meaningful exchanges between civil society leaders and other key decision-makers in the future of AI.

Perspectives from the Global South in general need to be centered in global discussions on AI governance. Besides Global South think tanks discussed earlier, greater representation from Global South governments—particularly regulators, technical representatives, and other public servants—would also be valuable in global discussions on AI governance, as they are often less equipped to respond to rapidly evolving technologies. In terms of safeguarding information integrity, it is increasingly important for the Global South and vulnerable groups (minorities, migrants, low-literacy, non-English-speaking, and Indigenous peoples) to be part of the efforts and considerations behind strategies against information pollution. Moreover, youth are particularly targeted for digital disinformation, through campaigns to manipulate their behaviours, beliefs, and voting preferences; hence, young people should be a central perspective in AI governance. Finally, a fundamental question is who is represented in the data that informs global discussions on AI governance? It is important to accelerate efforts to develop data ecosystems suitable to make AI work for everyone, including women and marginalised regions. Data must reflect the diversity of society, including regions, languages, informal work, and diverse gender realities. However, inclusion should not come at the cost of privacy and safety, which is why data protection and informed consent are crucial, and relevant frameworks should be built with the perspectives of women and human rights defenders.

The Dialogue can emphasise deliberate spaces where actors who don't often engage directly on questions of AI governance (such as governments, private sector, civil society, and technical communities) are brought together to confront difficult topics, including North-South power asymmetries and the responsibilities of AI developers. The format can embrace contested terrain. Sessions on private sector accountability, compute concentration, and extractivist dynamics in AI supply chains would benefit from structured debate formats that surface genuine disagreement. The expected outcomes of the Dialogue lend themselves well to this type of exchange. In addition, the lead up and follow up to this first Dialogue can function as a continuous process, not a one-off event. Building on models like this survey, a shared repository of inputs, recommendations, and responses — updated across the 2026 and 2027 sessions — would strengthen participation and ownership, relevance, and accountability.

Local data observatories: AI needs structured, reliable, and local data; in the context of a democracy, this also matters for increased transparency and to measure indicators for evidence-based policymaking. In Tunisia, the think tank Solidar Tunisie is implementing public policy monitoring tools that help make local realities visible. Through the Observatoire Tunisien des Politiques Publiques (OTPP), they are showing how local data can strengthen delivery and inclusion by tracking public investment and policy implementation. The public observatory measures what is being delivered, where gaps persist, and which regions are being left behind. UNESCO's Readiness Assessment Methodology (RAM): As stressed earlier in this survey, a practical pathway for the Global South towards more inclusive AI is to prioritise data readiness as much as AI readiness. UNESCO's Readiness Assessment Methodology (RAM) can help countries assess their overall preparedness, which can be complemented by looking at legal frameworks, stewardship, protection, and the ability to generate local datasets ethically. As it relates to the question of AI's role in the spread of misinformation, collaborations with fact-checkers, journalists, and tech platforms should also be expanded, given that they can help identify and counter false information swiftly.