Ministry of Foreign Affairs

A successful first Global Dialogue on AI Governance would deliver more than broad principles—it should create clear momentum, shared direction, and practical next steps. First, success would mean convergence on a common baseline of principles for AI governance. While full consensus is unlikely, agreement on core ideas such as human rights, transparency, accountability, and safety would establish a shared language across governments, companies, and civil society. This reduces fragmentation and helps align future regulatory efforts. Second, the dialogue should produce actionable pathways, not just declarations. This could include a roadmap for international cooperation, timelines for developing standards, or the creation of working groups on key issues like AI safety, data governance, and inequality. Without concrete follow-up mechanisms, the dialogue risks becoming symbolic. Third, it should meaningfully address global inequalities in AI development and access. A successful outcome would include commitments to capacity-building, technology transfer, and inclusive participation from the Global South, ensuring that governance frameworks are not dominated by a few countries or corporations. Fourth, success would involve multi-stakeholder legitimacy. Governments, private sector actors, academia, and civil society should all have visible influence in shaping outcomes. This builds trust and ensures that governance reflects diverse interests, not just state power or market priorities. Finally, the dialogue should establish continuity—for example, by institutionalizing the process through regular meetings or linking it to existing multilateral frameworks. One-off discussions are less impactful than sustained governance efforts. In short, success lies in moving from dialogue to direction: shared principles, concrete actions, inclusive participation, and a clear path forward.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Open-source software, open data and open AI models

Safe, secure and trustworthy AI is a top priority because the rapid deployment of AI systems without adequate safeguards poses real risks to individuals and societies. Advancing shared standards on safety, accountability, and transparency is essential to build trust and prevent harm across borders. AI capacity-building is equally urgent to ensure that all countries especially in the Global South-can meaningfully participate in the AI ecosystem. This includes strengthening technical skills, institutional capabilities, and access to infrastructure, so that governance is not dominated by a small group of actors. Interoperability of governance approaches is critical in a fragmented global landscape. With different national and regional regulations emerging, aligning frameworks where possible can reduce regulatory conflicts, facilitate international cooperation, and support innovation while maintaining safeguards. Together, these priorities reflect a balanced approach: mitigating risks (safety), enabling participation (capacity), and ensuring coordination (interoperability). This combination is key to building an inclusive and effective global AI governance system. Open-source software, open data and open AI models are essential for fostering transparency, innovation, and inclusive participation in the AI ecosystem. Open approaches can lower barriers to entry for researchers, startups, and governments especially in developing countries, supporting more equitable access to AI capabilities. They also enable greater scrutiny, which is key for improving safety, identifying biases, and building trust in AI systems. At the same time, advancing openness must be balanced with safeguards to prevent misuse, particularly for high-risk models. This makes it closely linked to your priority on safe, secure and trustworthy AI, as well as capacity-building, since countries need the skills and institutions to effectively use and govern open technologies.

First, compute and infrastructure governance is an increasingly critical gap. Access to high-performance computing (HPC) and cloud infrastructure is highly concentrated, shaping who can develop advanced AI systems. Without addressing this, both capacity-building and openness remain limited in practice. Second, data governance and data sovereignty require stronger emphasis. AI systems are deeply dependent on data flows that often cross borders, raising questions about ownership, consent, and fair value distribution-especially for countries in the Global South. This issue cuts across safety, interoperability, and equity. Third, AI and labor transformation is an urgent but underrepresented area. Beyond general "implications," there is a need for targeted global dialogue on job displacement, reskilling, and the future of work, particularly in developing economies where social protection systems may be weaker. Fourth, environmental sustainability of AI is emerging as a key concern. The energy and water consumption of large-scale AI systems can be significant, yet governance discussions rarely integrate environmental standards or reporting requirements. Finally, governance of frontier AI models-including questions around release strategies and international oversightre mains insufficiently defined. This intersects with safety, open models, and interoperability, but may require more specialized mechanisms. Highlighting these issues would strengthen the dialogue by addressing the structural drivers of inequality, sustainability, and long-term risk in the AI ecosystem.

Governance gaps in safe, secure and trustworthy AI, AI capacity-building, interoperability, and open AI ecosystems are already affecting Colombia and Latin America, particularly from the perspective of the Ministry of Foreign Affairs. A central challenge is the limited understanding and policy clarity around digital sovereignty and AI regulation. In Colombia, there is still a lack of consensus on what concrete measures should be adopted to govern AI, protect data, and ensure national control over critical digital infrastructures. This creates uncertainty for both domestic policy and international positioning. A second major constraint is the lack of digital infrastructure. Limited access to high-performance computing, data centers, and advanced connectivity restricts the country's ability to develop, deploy, and govern AI systems effectively. This reinforces dependency on external technologies and reduces strategic autonomy. Additionally, insufficient technical and institutional capacity affects Colombia's ability to engage in global AI governance discussions. This can lead to asymmetric participation in multilateral negotiations, where countries in the region risk becoming rule-takers rather than active contributors. At the same time, there are important opportunities. Open-source AI, open data, and international cooperation can help reduce barriers to entry and strengthen local innovation ecosystems. For the Ministry of Foreign Affairs, this opens space to promote regional collaboration and position Colombia as a constructive actor advocating for inclusive and interoperable governance frameworks. Overall, these gaps highlight a dual dynamic: while Colombia faces structural limitations in regulation, knowledge, and infrastructure, it also has an opportunity to leverage diplomacy and openness to strengthen its role in shaping a more equitable global AI governance system.

The AI Dialogue can play a critical role as a neutral, multilateral platform that fosters convergence in an increasingly fragmented global AI governance landscape. At a time when different regulatory approaches are emerging across regions, the Dialogue can help build shared understanding and trust among states, the private sector, and other stakeholders. One of its key contributions is enabling policy coordination without requiring full harmonization. By promoting interoperability of governance approaches, the Dialogue can reduce regulatory conflicts and facilitate cross-border collaboration while respecting national contexts. The Dialogue can also serve as a space to elevate the voices of developing countries, including Colombia and Latin America, ensuring that global AI governance is not shaped exclusively by a few major powers. This is particularly important for advancing capacity-building, knowledge-sharing, and more equitable participation in decision-making processes. Additionally, it can help translate high-level principles into practical cooperation mechanisms, such as joint initiatives, technical working groups, and shared standards on AI safety and transparency. Finally, the AI Dialogue can strengthen the link between AI governance and broader multilateral agendas, including sustainable development, human rights, and digital inclusion, reinforcing a more holistic and inclusive approach to global governance.

The AI Dialogue should build upon existing international and multistakeholder efforts such as the work of the United Nations system, including UNESCO's Recommendation on the Ethics of AI, as well as initiatives like the OECD AI Principles and the Global Partnership on AI (GPAI). Regional efforts in Latin America and ongoing discussions in forums such as the G20 also provide important foundations. These initiatives have already advanced norm-setting, ethical frameworks, and technical cooperation, but they often operate in parallel and with limited coordination. The added value of the AI Dialogue lies in its ability to act as a convening and bridging mechanism. It can connect fragmented efforts, promote coherence across frameworks, and facilitate dialogue between regions with different regulatory traditions. Importantly, the Dialogue can also fill gaps by focusing on implementation and inclusivity. While many existing initiatives define principles, fewer provide pathways for countries—especially in the Global South—to operationalize them. The Dialogue can support this through capacity-building, peer learning, and the exchange of best practices. In this sense, its value is not to duplicate existing efforts, but to align, amplify, and operationalize them within a truly global and inclusive platform.

Different stakeholders can contribute by bringing complementary perspectives that reflect the complexity of AI governance. States should lead on defining policy priorities, ensuring alignment with international law, and representing national and regional interests. They can also help translate global discussions into regulatory frameworks. International organizations can provide technical expertise, facilitate coordination, and ensure continuity between the Dialogue and existing multilateral efforts. They are key to maintaining neutrality and institutional memory. Private sector actors should contribute practical insights on technological development, risks, and implementation challenges. Their participation is essential for ensuring that governance frameworks are realistic and operational. Academia and civil society play a critical role in ensuring evidence-based discussions, ethical scrutiny, and accountability, particularly regarding human rights and social impacts. In terms of structure, the AI Dialogue should adopt a hybrid model combining: High-level ministerial segments for political direction Technical working groups for specific issues (e.g., safety, capacity-building) Multi-stakeholder roundtables for inclusive exchange Regional consultations to ensure geographic balance

Underrepresented voices and how to include them Currently underrepresented voices include Global South countries, especially smaller developing states with limited technical capacity, as well as local communities affected by AI deployment, indigenous groups, and workers in sectors undergoing AI-driven transformation. To include them, the Dialogue should ensure funded participation mechanisms, remote access options, and simplified technical formats that reduce entry barriers. Regional consultations in Latin America, Africa, and Asia should feed directly into global discussions. There is also a need to better include non-English-speaking stakeholders and actors outside major tech hubs. Translation services, capacity-building workshops, and partnerships with regional organizations can help bridge this gap.

Innovative engagement formats for meaningful participation The AI Dialogue could benefit from more interactive and inclusive formats beyond traditional panel discussions. Examples include: Policy labs or scenario workshops, where stakeholders jointly test governance approaches Simulation exercises, exploring real-world AI risks and responses Multi-stakeholder drafting sessions, producing co-authored policy recommendations Digital participation platforms, allowing asynchronous input from global participants Regional "listening sessions", feeding structured inputs into global meetings These formats would make engagement more participatory, evidence-based, and action-oriented, moving beyond dialogue toward co-creation of governance solutions.

At the ethical and regulatory level, Colombia has advanced through CONPES 4144 (2025), which explicitly promotes the use of algorithmic impact assessments in the public sector, the adoption of human-in-the-loop principles for automated decision-making systems, and the integration of data governance standards aligned with constitutional protections such as habeas data. In practice, this is relevant for systems used in areas like citizen complaint management platforms and early risk-scoring tools in public service delivery, where human oversight is required before decisions are finalized. Additionally, the Ministry of Information and Communications Technologies (MinTIC) has implemented digital government initiatives such as the Interoperability Framework (Interoperabilidad de Datos Públicos), which allows different state institutions (e.g., registries, tax authority, and social programs) to exchange structured data. This infrastructure is a foundational layer for future AI systems in the public sector, although coverage remains uneven outside major cities. MinTIC has also promoted GovTech and innovation labs (Labs de Innovación Pública Digital), where pilot projects test data analytics and AI tools-for example, chatbots for citizen services and predictive analytics for administrative processes. However, these pilots often depend on external vendors and cloud providers, limiting internal technological sovereignty. On the international side, Colombia participates in OECD digital transformation initiatives and regional digital government cooperation platforms in Latin America, which facilitate knowledge exchange on AI governance and digital public infrastructure. Despite this, the country still faces structural constraints in high-performance computing access, specialized AI talent pipelines, and domestic cloud/data infrastructure, which limits the scalability of AI deployment. Overall, while Colombia has moved from principles to early implementation through concrete instruments and pilot systems, the main challenge remains the transition toward institutionalized, scalable, and infrastructure-backed AI governance.