MKAS Group

• The first Global Dialogue on AI Governance would be a success if it led to concrete and inclusive commitments. This would mean: the adoption of a shared ethical framework for AI development and deployment, recognized by both developed and developing nations
• the establishment of transparent monitoring mechanisms to assess compliance by public and private actors
• and the meaningful inclusion of developing countries in decision-making processes, ensuring AI governance is not dominated solely by major tech powers. Success would also require commitments on sharing the benefits of AI to bridge digital inequalities. Finally, agreements on preventing risks from high-stakes AI systems in military, health, or judicial domains would send a strong signal of collective willingness to act responsibly.

• 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

he four selected domains represent the essential foundations of a truly responsible and human-centered AI governance framework. Safe, secure, and trustworthy AI is an absolute priority, as the large-scale deployment of AI systems without adequate safeguards exposes individuals and societies to significant risks misinformation, cyberattacks, algorithmic bias, and unanticipated job displacement. Trust is the cornerstone of any beneficial AI adoption. The social, economic, ethical, cultural, linguistic, and technical implications of AI form a crucial cross-cutting domain. AI does not operate in a vacuum: it transforms labor markets, amplifies or reduces inequalities, and can marginalize entire cultures and languages if not developed inclusively. Effective governance must account for this multidimensional complexity. The protection and promotion of human rights are non-negotiable. AI can threaten privacy, freedom of expression, equality before the law, and human dignity if deployed without appropriate legal frameworks. Human rights must be at the core not the periphery of any AI governance regime. Transparency, accountability, and human oversight are concrete mechanisms to correct abuses. Without transparency on how algorithms function and clear accountability structures, it is impossible to ensure that AI serves the public interest. Human oversight remains essential to prevent the blind automation of decisions that affect human lives. These four priorities form a coherent and complementary set, essential for ensuring that the Global Dialogue leads to truly transformative commitments.

Yes, several cross-cutting and emerging issues deserve to be integrated into the Global Dialogue on AI Governance, as they are not sufficiently covered by the themes identified so far. The environmental impact of AI is a major and often overlooked challenge. Training large AI models consumes massive amounts of energy and water, contributing significantly to greenhouse gas emissions. Responsible governance must incorporate environmental sustainability requirements for AI systems. The digital divide and equitable access represent a critical challenge, particularly for countries in the Global South. If the benefits of AI remain concentrated in a few technologically advanced nations, global inequalities risk deepening further. Governance must ensure fair access to AI technologies and their benefits. AI and democracy is an urgent emerging issue. Recommendation algorithms, deepfakes, and automated disinformation threaten the integrity of electoral processes and social cohesion. Specific mechanisms to protect democracies against malicious uses of AI must be developed. The concentration of power among a few private actors is also concerning. A small number of companies currently controls most of the infrastructure, data, and most powerful AI models. This concentration creates risks of technological dependency for states and weakens the digital sovereignty of nations. Finally, AI in conflict and military contexts remains insufficiently regulated. The development of lethal autonomous weapons raises fundamental ethical and legal questions that require an urgent and coordinated international response. These issues must be explicitly included in the Dialogue's agenda to ensure its relevance and comprehensiveness.

In West Africa, and particularly in Togo, gaps in AI governance manifest in concrete and multidimensional ways, generating both significant challenges and opportunities to be seized. Challenges: The absence of national regulatory frameworks adapted to AI leaves citizens and institutions vulnerable to risks associated with algorithmic systems particularly regarding personal data protection, bias in automated decisions, and disinformation. Insufficient digital infrastructure slows equitable AI adoption and deepens the digital divide between urban and rural areas. Furthermore, the lack of local technical expertise researchers, engineers, and legal specialists in AI limits the country's ability to actively participate in global discussions and develop solutions tailored to local realities. Cultural and linguistic implications are also concerning: most large AI models ignore African languages, threatening cultural diversity and the accessibility of digital tools for local populations. Opportunities: Despite these challenges, AI represents a considerable opportunity to accelerate development. In health, agriculture, education, and financial services, AI solutions adapted to local contexts could transform the lives of millions of people. Togo and the broader region possess a dynamic youth population and a rapidly growing digital ecosystem, capable of leveraging these technologies if a supportive governance framework is established. The Global Dialogue therefore represents a unique opportunity for African countries to voice their specific needs, secure support for capacity building, and contribute to shaping a truly global and inclusive AI governance architecture one that reflects the realities and aspirations of all nations, not just the most technologically advanced ones.

The AI Dialogue can play a central and structuring role in promoting international cooperation on AI governance at several levels. A space for multilateral convergence. The Dialogue offers a unique platform to bring together currently fragmented regulatory approaches between the European Union, the United States, China, and the Global South in order to identify common principles and prevent global digital fragmentation. It can foster the emergence of shared language and minimum common standards. A trust-building mechanism between nations. By bringing governments, civil society, the private sector, and the scientific community together, the Dialogue can establish the climate of trust essential to any lasting cooperation. It can also serve as a framework for transparent exchanges on the capabilities, risks, and intentions of different AI actors. A lever for the inclusion of developing countries. The Dialogue can play a redistributive role by facilitating knowledge transfer, capacity building, and access to technological resources for less advanced nations. It can thereby correct current asymmetries and ensure that global AI governance genuinely reflects the diversity of contexts and needs. A catalyst for collective action. Faced with cross-border challenges disinformation, cyber threats, autonomous weapons no country can act alone. The Dialogue can coordinate collective responses and establish operational cooperation mechanisms, particularly for managing AI-related crises. In short, the AI Dialogue can be far more than a consultative forum: it can become the foundation of a robust, legitimate, and inclusive international cooperation architecture.

The AI Dialogue should build on several existing initiatives and mechanisms, while bringing its own distinct added value. Existing initiatives to mobilize: The UNESCO Recommendation on the Ethics of AI (2021) provides a solid ethical foundation, already adopted by 193 member states, upon which the Dialogue can build to operationalize common principles. The OECD AI Principles offer a recognized reference framework for member countries and beyond. The Global Digital Compact of the United Nations, adopted in 2024, provides a relevant multilateral engagement framework. At the regional level, the EU AI Act represents a concrete regulatory experience whose lessons can be shared. The African Union has also adopted a continental AI policy, offering an indispensable Southern perspective. Finally, forums such as the Global Partnership on AI (GPAI) and ITU AI for Good are already active technical and multi-stakeholder spaces. The added value of the AI Dialogue: Where these initiatives often remain sectoral, non-binding, or limited to certain regions, the AI Dialogue brings a truly universal and intergovernmental dimension. It can bridge these fragmented mechanisms, avoid duplication, and promote coherence among them. Its added value also lies in its ability to integrate voices often absent from major negotiations civil societies from the Global South, indigenous peoples, small island states and to translate principles into concrete, measurable political commitments. It can also serve as a periodic review mechanism, adapting global governance to the rapid pace of technological advancements.

For the AI Dialogue to be truly inclusive and representative, it is essential to design a structured multi-stakeholder format allowing all parties to contribute meaningfully. Governments should participate through official national delegations, with balanced representation between developed and developing countries. Regional quotas could ensure that Africa, South Asia, Latin America, and small island states have a genuine voice in negotiations. Civil society should be integrated not as passive observers, but as full participants. Dedicated sessions, formal speaking rights, and prior consultation mechanisms would help reflect the concerns of ordinary citizens, marginalized communities, and rights advocacy organizations. The private sector, particularly technology companies, must contribute transparently, with mandatory conflict-of-interest declarations to prevent regulatory capture. Their technical expertise is valuable, but their influence must be appropriately regulated. The scientific and academic community should inform debates with evidence-based data, independent risk assessments, and research-driven proposals. Regarding structure, the Dialogue should adopt a hybrid format in-person and online to maximize global participation, particularly from low-resource countries. Regional preparatory meetings would help consolidate positions before plenary sessions. A permanent secretariat would ensure continuity between sessions. Finally, clear follow-up mechanisms with implementation indicators would transform discussions into concrete and accountable commitments.

Several voices, communities, and perspectives remain largely underrepresented in global AI governance discussions, constituting a major democratic deficit. Countries of the Global South are the primary victims of this marginalization. Sub-Saharan Africa, Southeast Asia, and Latin America are home to the majority of the world's population, yet their governments, researchers, and civil society are nearly absent from major AI negotiations. Their realities informal economies, minority languages, limited infrastructure are not reflected in current normative frameworks. Indigenous communities are also absent, despite being directly affected by AI applications related to surveillance, natural resource management, and facial recognition. Their knowledge systems and collective rights must be taken into account. Women and gender minority groups remain underrepresented in both decision-making spaces and technology development itself, even though algorithmic biases affect them disproportionately. Small island states and least developed countries lack the resources to actively participate in international negotiations and to develop national positions on AI. To include them, several concrete measures are needed: funding for the participation of Southern delegations; creation of regional and local consultation platforms prior to global negotiations; integration of representation quotas in the Dialogue's structure; capacity-building support so these actors can engage in an informed manner; and formal recognition of contributions from indigenous communities and grassroots organizations in final documents.

To foster meaningful and dynamic engagement during the AI Dialogue, innovative and participatory formats must be prioritized, breaking away from traditional conference models that are often too formal and insufficiently interactive. Structured deliberative sessions inspired by the citizens' assembly model would allow participants from diverse backgrounds to debate in small mixed groups, formulate consensus recommendations, and present them in plenary. This format encourages mutual listening and the co-construction of solutions. Policy innovation labs bringing together technical experts, policymakers, and civil society representatives around concrete cases and simulations would allow regulatory approaches to be tested before adoption. These prototyping spaces encourage a pragmatic, solution-oriented approach. Inclusive digital dialogues, combining online participation platforms, asynchronous consultations, and automatic translation tools, would allow participants from around the world to engage including those who cannot travel physically for financial or logistical reasons. Intergenerational panels associating young innovators, emerging researchers, and experienced decision-makers would enrich debates with long-term perspectives and visions grounded in current technological realities. Structured informal networking spaces thematic speed-networking sessions, co-drafting workshops, governance hackathons would foster the creation of concrete partnerships between participants. Finally, a real-time synthesis mechanism, with regular summaries of debates accessible to all participants, would guarantee transparency and allow everyone to follow the evolution of discussions and contribute to them continuously and in an informed manner.

Several policies, practices, and platforms constitute inspiring examples of effective AI governance, from which the Global Dialogue could draw lessons. At the regulatory level, the EU AI Act, adopted in 2024, represents the first comprehensive AI legislation. Its risk-based approach prohibiting certain dangerous applications and imposing strict obligations on high-risk systems offers a concrete and adaptable regulatory model. At the ethical level, the UNESCO Recommendation on the Ethics of AI (2021) provides a universal normative framework covering human rights, environmental sustainability, and inclusion. Its Readiness Assessment Methodology the RAM helps countries measure their governance capacities. At the technical level, the UK's AI Safety Institute and its American counterpart have developed methodologies for evaluating frontier AI models, particularly to detect dangerous capabilities. These independent auditing approaches are essential for credible oversight. At the African continental level, the African Union Policy on AI (2024) represents a remarkable initiative adapted to the continent's realities, emphasizing inclusive development, data sovereignty, and local capacity building. At the multi-stakeholder platform level, the Global Partnership on AI (GPAI) brings together governments, researchers, and civil society around thematic working groups producing practical recommendations on topics such as labor, health, and data governance. Finally, initiatives like ITU's AI for Good demonstrate how AI can be mobilized to achieve the Sustainable Development Goals, offering concrete use cases in health, agriculture, and climate change particularly relevant for developing nations seeking to harness AI's potential responsibly.