Fundación GeoCensos

A successful inaugural Global Dialogue must transcend abstract ethical principles to address the high-stakes, "ground-truth" realities of Artificial Intelligence. Success is defined by the recognition of Geospatial AI (GeoAI) as a critical geopolitical priority, requiring a governance framework that is as dynamic and precise as the territories it models. A relevant measure of success should be the adoption of a Governance Integrated Layer (GIL). Rather than creating institutional overlap, the GIL acts as a transversal framework that aligns the mandates of the ITU, UNESCO, UNOCHA, and the World Bank across the entire GeoAI stack. This ensures that the "Data Revolution" does not become an "Accountability Gap" where models trained in the Global North misinterpret the informal economies or dynamic land use of the Global South. Tangible deliverables making this Dialogue a success include: 1. A Roadmap for a Spatial Accountability Board: Establishing mechanisms to validate GeoAI systems that influence public decision-making, ensuring that "diffuse responsibility" is replaced by "territorial accountability." 2. The Geographic Transferability Framework: A technical and policy protocol to ensure AI models are context-aware, adapted, and tested for the specific socio-environmental realities of the regions they enter. 3. Principles for Territorial Data Sovereignty: Explicit guidelines that empower countries to manage their local data ecosystems, moving from being sources of raw data to producers of sovereign knowledge. Ultimately, success occurs when the Dialogue transitions from "AI in the cloud" to "AI in the territory." By embedding the GIL, the international community ensures that technological scaling does not result in "spatial misreading," but rather in inclusive, regionally relevant, and verifiable intelligence that respects the lived experience of every community. Success is a coordinated oversight system where AI is not just technically sound, but territorially fair.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

The selected priorities reflect the urgent need to address the systemic gaps that uniquely characterize GeoAI governance. Safe, secure and trustworthy AI is foundational to GeoAI systems as they influence high-stakes entirely involving territorial decisions. Unlike other AI domains, errors in spatial modeling can result in misallocated vital resources for local communities, overlooking environmental policies on the ground, or inadequate disaster responses to aid vulnerable populations. Trustworthiness must therefore include spatial, human - driven validation, not only technical robustness. AI capacity-building is particularly relevant to assure access to Geo AI technologies in all countries, especially those in the Global South, where massive amounts of qualitative geospatial data are being generated from official and alternative sources, but where technical and institutional capabilities often lag. Capacity-building should not just develop skills but also empower institutions in emerging countries to engage in governance, local validation of models, and sustain data sovereignty. Interoperability of governance approaches is at the core of this integrative input, as it directly addresses the fragmented nature of current oversight of AI operations. GeoAI spans multiple layers of components such as earth observation sensors, data standards, cloud infrastructure, and AI models. At present, each of these parts is governed or supervised by different supranational institutions. Without interoperability, governance risks to remain siloed, undermining coherent policy implementation across borders. Transparency, accountability, and human oversight are critical to closing gaps in AI governance. GeoAI operates across complex institutional arrangements where responsibility is often diffuse. Independent audits and registries for GeoAI systems are crucial for making territorial decisions transparent and fit for purposes. The above priorities enable a GIL that ensures GeoAI systems function responsibly across regions, addressing accountability, transferability, and infrastructure sovereignty gaps.

A critical cross-cutting issue not fully captured by the listed themes is the spatial dimension of AI governance for the geospatial space, which introduces challenges that are both technical and geopolitical. The transferability of AI models (formulated mostly by Global North academia, practitioners and think tanks) remains largely unaddressed. All AI systems are inherently context-dependent, yet models are frequently deployed across regions without adequate adaptation. In the case of Geo AI, this creates risks of geographic misrepresentation, particularly when models developed in other regions are applied in the Global South without validation. Governance frameworks must therefore incorporate mandatory spatial testing and certification protocols. Another emerging concern is the infrastructure and sovereignty dimension. In the case of GeoAI, it operates within a layered architecture where data collection, processing, and model deployment are often geographically and institutionally disjointed. This asymmetry raises fundamental questions about who controls data, who benefits from its analysis, and under which jurisdiction governance applies. Current frameworks insufficiently address these power imbalances. These issues point to the need for integrating approaches to introduce transferability standards, assess infrastructure asymmetries, and safeguard territorial data sovereignty. Without addressing these dimensions, global AI governance risks overlooking how algorithms reshape the physical world, leaving critical decisions about land, resources, and infrastructure insufficiently governed.

As we accelerate toward the 2030 Agenda, the integration of Geospatial Artificial Intelligence (GeoAI) in regions such as Latin America, Southeast Asia, and the African continent present a dual reality. While these technologies offer unprecedented potential for the SDGs, significant governance voids threaten to entrench existing inequalities. The growing reliance of the public sector on externally developed geospatial models, including but not limited to land-use classification, disaster response, and urban planning, has created a crisis of responsibility. When proprietary algorithms produce biased or flawed outputs, the lack of calibration frameworks weakens institutional trust. The spatial transferability gap is prominent in the Global South; models trained in temperate or urban settings often perform poorly in tropical, peri-urban, or less developed areas. This gap can lead to misrepresented agricultural policies, flawed risk assessments, and inequitable infrastructure planning, leaving behind numerous vulnerable populations. Global South is often viewed mainly as a supplier of raw data, while most data management and infrastructure are controlled elsewhere. This limits national control over data and decision-making. Enhancing Data Sovereignty is essential for lasting political and economic fairness. Defragmenting means in this case building a GIL with all relevant stakeholders on the ground, i.e. a framework that turns systemic weaknesses into institutional strengths. By decentralizing modelling, this method encourages GeoAI systems tailored to Global South's specific needs, improving local accuracy and bridging gaps in SDG data reliability. Implementing an integrated model fundamentally redefines the relationship between technology and territory by championing Data Sovereignty. By investing in localized computational infrastructures and managerial expertise, nations can reclaim their role as producers of knowledge rather than mere sources of raw data. This structural evolution provides the necessary autonomy to manage territorial realities and ensures that accountability is in place.

The AI Dialogue can play a catalytic role in advancing international cooperation by serving as a convergence platform for aligning fragmented governance efforts. Strengthening constructive AI Dialogue over time could help establish GeoAI as a distinct field for international cooperation across countries and territories, especially among Global South members. Coordinated oversight encourages shared understanding that territorial AI needs spatially grounded governance. This approach can connect data, infrastructure, and ethics-focused institutions more effectively. The Dialogue could facilitate the co-creation of an Integrated Layer, enabling interoperability between existing frameworks while avoiding institutional duplication. This would enhance cooperation by establishing common reference points for accountability, transferability, and sovereignty. Additionally, the Dialogue can promote trust-building measures, such as shared frameworks and standards for spatial validation and transparent registries of GeoAI models, and applications of global relevance. These mechanisms would support cross-border collaboration while addressing concerns about data misuse and unequal power dynamics. Importantly, the Dialogue provides an opportunity to strengthen the voice of the Global South in shaping governance norms. By fostering inclusive participation, it can ensure that international cooperation reflects diverse territorial realities and avoids one-size-fits-all approaches. In essence, the AI Dialogue can transform international cooperation from a collection of parallel initiatives into a cohesive governance ecosystem, where GeoAI systems are developed, validated, and deployed in ways that are accountable, interoperable, and responsive to the needs of all regions.

The AI Dialogue should anchor its efforts in established institutional foundations of existing AI related ecosystems while actively incorporating global initiatives that have successfully advanced data availability, strengthening mechanisms for integrated oversight across the artificial intelligence lifecycle. The central mechanism of this proposal is the GIL, presented above, which serves as a cross-cutting regulatory and operational framework designed to harmonize the fragmented oversight of artificial intelligence applied to the geospatial realm. The GIL acts as a coordinated "connective tissue" that aligns the various components of the GeoAI architecture. Its layout integrates a wide variety of elements, ranging from satellite data acquisition and cloud infrastructure to the implementation of algorithms, all overseen within a unified accountability framework. Another starting point for the Global Dialogue on AI is the ISO 42001 standard, which provides a certifiable framework for AI management systems, ensuring that technical soundness is accompanied by organizational accountability. The Center for AI and Digital Policy (CAIDP) can help further strengthen the Geospatial Dialogue on AI Governance. This organization offers a highly useful model through its "AI Policy Clinics." These clinics provide the rigorous research and training needed to evaluate national AI strategies in light of democratic values. This is crucial for the proposed GIL to ensure "geospatial accountability," which is not merely a technical indicator but a priority regarding sovereign rights. This is particularly urgent given that, at present, technical experts and local authorities from emerging economies remain significantly underrepresented in these global standard-setting processes. By uniting fragmented AI oversight, ISO standards, and CAIDP's policy reviews, the AI Dialogue can evolve from scattered efforts to a unified, reliable ecosystem. This alignment ensures that GeoAI is not merely "technically sound" but "territorially equitable," providing a clear roadmap for the Spatial Accountability Board and Geographic Transferability Group to address critical governance gaps effectively.

A wide array of stakeholders can contribute to the AI Dialogue in the Geospatial realm by aligning their expertise with specific layers of the GeoAI governance ecosystem. Governments should provide policy direction and regulatory leadership, particularly in defining accountability frameworks and safeguarding data sovereignty. International organizations can facilitate normativity alignments and coordination, ensuring interoperability across jurisdictions. The private sector, including cloud and satellite providers, should contribute with technical expertise, transparency in algorithms, and technological transfer, particularly regarding infrastructure and model deployment. Academia and civil society can play a critical role in on-the-ground validation, auditing, and advocacy, ensuring that governance remains evidence-based and inclusive. To maximize effectiveness, the Dialogue should create a space for GeoAI Governance to link sector discussions and maintain continuity, hosting forums such as a Global Geospatial Governance Forum for ongoing collaboration. Iterative mechanisms should be included, so that stakeholders can confirm, follow up and update governance frameworks responsive to technological and geopolitical changes. By structuring participation around the GeoAI stack and fostering continuous engagement of relevant stakeholders, the Dialogue can move beyond declarative outcomes toward operational governance solutions that are both inclusive and actionable.

Global discussions on AI governance, most probably due to the nature of rapid technological advances, continue to underrepresent several critical voices, in particular those directly affected by GeoAI systems on the ground. Local and subnational authorities in the Global South are alarmingly unaware of these vital dialogues, often being paradoxically those responsible for implementing policies informed by GeoAI but have limited influence over how these systems are designed and governed. Their exclusion intensifies the spatial transferability gap, as models fail to reflect local realities, especially when data are not updated and sovereign considerations are disregarded. Beyond the technical possibilities, indigenous and community-based stakeholders are commonly underrepresented, despite their deep knowledge of territorial dynamics and their direct stake in land-use decisions. Their inclusion is essential for ensuring that GeoAI systems respect cultural and environmental contexts. The role of technical experts, especially from emerging economies and territories, remains underrepresented in global standard-setting processes. This seriously limits the diversity of perspectives in developing frameworks for any kind of comprehensive governance and reinforces existing asymmetries in infrastructure and knowledge production. To address these gaps, the AI Dialogue for the geospatial and many other realms should implement targeted inclusion mechanisms, such as regional consultations, funded participation programs, and multilingual engagement platforms. It should also promote community-in-the-loop governance mechanisms, ensuring that local stakeholders are actively involved in validation and oversight processes. Embedding these perspectives within a Governance Integrated Layer would enhance legitimacy, improve model performance across diverse geographies, and ensure that GeoAI governance reflects the realities of all regions, amplifying the territorial outreach of advanced technological capabilities.

To enable effective engagement, the AI Dialogue should use new formats if it intends to address the complex and interdisciplinary aspects of AI governance. In the case of GeoAI applications an effective approach would be in person and on line scenario-based simulations, where participants collaboratively prioritize the governance implications of real-world, such as disaster response or urban planning. These exercises can reveal accountability gaps and highlight the need for spatial validation in a practical context. Another format is organizing multi-stakeholder labs, bringing together policymakers, technical experts, and community representatives to co-design governance solutions. These labs could focus on specific challenges, such as developing transferability protocols or defining data sovereignty frameworks. The Dialogue could also introduce AI audit demos, showcasing how independent validation processes can be conducted across different regions. This would provide tangible insights into the operationalization of accountability mechanisms. Aided by balanced partnership arrangements with the private sector, d digital platforms should support sustainable engagement of participants of the Dialogue, enabling stakeholders to contribute beyond formal sessions. Interactive registries or dashboards could allow participants to explore diverse AI applications and governance practices in real time. Finally, planning for future editions of the AI Dialogue, inputs from regional dialogue clusters can ensure that discussions are grounded in local contexts while contributing to global outcomes. These clusters would feed into the central Dialogue, creating a feedback loop between global principles and territorial realities. By combining these and other innovative formats, the AI Dialogue can move from abstract discussions to experiential and solution-oriented engagement, fostering with the passing of time a deeper understanding of GeoAI governance challenges and enabling collaborative innovation.

Emerging GeoAI practices offer useful insights for governance, though the field is fragmented. The AI Dialogue should focus on addressing this fragmentation with a practical, holistic approach. While ethical guidelines provide a foundation, their effectiveness is limited without enforcement tools, especially in spatial contexts. One promising approach is the development of data-sharing frameworks that promote open access to Earth observation data while maintaining quality standards. These frameworks demonstrate the potential for global collaboration but need to be complemented by mechanisms ensuring that data is used responsibly and contextually. To ensure these frameworks are "context-aware," the "senti-pensante" (feeling-thinking) participatory methodology practiced in Colombia should also be considered. This approach bridges the gap between technical data and lived territorial experience, ensuring that GeoAI systems are validated by the communities they impact. Another emerging model is the use of registries for AI systems, like the EU Database for High-Risk AI Systems under the EU AI Act, boost transparency by requiring providers to document applications before deployment. Using this approach in GeoAI could enhance traceability and accountability in critical territorial models. Building on these and other valuable pilots, the proposed Governance Integrated Layer offers a comprehensive solution. By combining a Spatial Accountability Board, a Geographic Transferability Group, and mechanisms for data sovereignty and transparency, it provides a structured approach to addressing the challenges of GeoAI. The practices depicted illustrate the underlying intent of this input for effective governance of AI in the geospatial realm. The real challenge isn't creating brand-new systems. It is about harnessing and expanding proven strategies so that GeoAI drives accountability, equity, and adapts effectively to the dynamic realities of our physical world. It's about uniting what works and pushing it further, ensuring GeoAI not only meets today's demands but shapes a future that's fair and transparent for all.