Saint Clair Strategies

The first Global Dialogue on AI Governance will be a success if it produces three tangible outcomes. First, a shared framework for inclusive participation in ongoing AI governance processes, one that moves beyond token representation of civil society and developing country stakeholders toward structural roles in agenda-setting, deliberation, and follow-up. The Dialogue's legitimacy depends on whether the communities most affected by AI can genuinely shape it, not merely observe it. Second, concrete commitments on capacity-building that address the regulatory asymmetry facing smaller economies. When major jurisdictions publish AI frameworks independently, the compliance burden falls disproportionately on least-resourced countries. The Dialogue should produce actionable mechanisms for regulatory cooperation and technical assistance particularly for Latin American and Caribbean governments navigating simultaneous pressure from U.S., EU, and emerging regional frameworks. Third, a clear articulation of human rights as a structural requirement not a thematic annex running through every work stream. In the LAC region, algorithmic systems are already shaping access to credit, immigration status, and public benefits with limited transparency, recourse, or accountability. The Dialogue must produce language and commitments that translate human rights principles into enforceable governance standards. Beyond these substantive outcomes, success also requires a process commitment: that the Independent International Scientific Panel's inaugural report is not simply presented to the Dialogue but genuinely informs its thematic discussions and policy recommendations. Finally, linguistic inclusion must be operationalized. AI systems trained predominantly on English-language data produce material harms for Spanish, Portuguese, and Indigenous-language speakers. Capacity-building investments must include multilingual data infrastructure not governance literacy alone.

• Protection and promotion of human rights
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• AI capacity-building

These four areas reflect the most urgent governance gaps facing Latin American and Caribbean communities and smaller economies navigating the current AI landscape. Human rights protection is the foundation. In the LAC region, algorithmic systems are already determining access to credit, immigration status, and public benefits often with no transparency, recourse, or accountability framework. Without human rights as a structural requirement rather than an aspirational principle, governance frameworks risk legitimizing harm. The social, economic, ethical, cultural, and linguistic implications of AI are acutely felt in communities where AI systems are built on data that does not reflect their languages, cultures, or lived realities. Spanish, Portuguese, and Indigenous-language speakers face compounding disadvantages when AI systems are trained predominantly on English-language data. These are not peripheral concerns they are material harms requiring dedicated attention. Interoperability of governance approaches is a political as much as a technical challenge. Smaller economies cannot bear the compliance burden of simultaneously adapting to U.S., EU, Brazilian, and other national AI frameworks. The Dialogue must prioritize practical regulatory cooperation mechanisms that do not export that burden to least-resourced countries. AI capacity-building is the prerequisite for meaningful participation in all other areas. Countries that lack the technical infrastructure, regulatory expertise, and multilingual data resources to govern AI cannot be expected to shape global norms on equal footing. Capacity-building must extend beyond governance literacy to include investment in high-performance computing, multilingual datasets, and local regulatory capacity.

Two cross-cutting issues warrant explicit attention in the Dialogue's structure. First, the intersection of AI governance and cybersecurity. As AI systems become embedded in critical infrastructure energy grids, financial systems, public health platforms the governance frameworks that address safety, accountability, and transparency must be integrated with existing cybersecurity frameworks rather than developed in parallel silos. The LAC region in particular faces compounding vulnerabilities where weak cybersecurity postures intersect with rapid AI deployment and limited regulatory capacity. The Dialogue should explore how existing frameworks such as NIST CSF and regional cybersecurity agreements can be leveraged and adapted for AI governance. Second, the governance of AI in the context of migration and cross-border data flows. AI systems are increasingly used in immigration enforcement, asylum processing, and border management contexts where the stakes for individuals are existential, and the accountability gaps are severe. This issue cuts across human rights, data governance, and interoperability in ways that none of the listed themes fully captures on its own. The LAC region, with significant populations affected by migration-related AI deployments both within the hemisphere and at its borders, has a particular stake in ensuring this issue receives dedicated attention. Both issues reflect a broader gap in the current thematic structure: the absence of an explicit focus on AI governance in high-stakes, life-affecting decision-making contexts where the consequences of ungoverned AI are most severe and the affected communities are least represented in governance processes

The Latin American and Caribbean region sits at a critical inflection point in AI governance. Significant AI deployment is already underway across the region in financial services, public administration, healthcare, and law enforcement while the regulatory frameworks, technical capacity, and institutional infrastructure needed to govern these systems remain nascent or absent in most countries. The most significant challenge is asymmetric exposure. LAC communities are subject to AI systems they did not design, trained on data that does not reflect their languages or contexts, governed by frameworks developed elsewhere, and with limited ability to contest or appeal AI-driven decisions that affect their lives. This asymmetry is compounded by the regulatory fragmentation challenge: as the United States, European Union, and Brazil each advance distinct AI governance frameworks, smaller LAC economies face impossible compliance burdens with limited technical assistance. Human rights gaps are acute and documented. Algorithmic systems used in credit scoring, immigration processing, and social benefit determination have produced discriminatory outcomes affecting vulnerable populations across the region. Redress mechanisms are limited. Transparency requirements are weak or unenforced. Civil society organizations working to document and challenge these harms operate with minimal resources. At the same time, the LAC region presents genuine opportunities. Brazil's AI Act, Colombia's emerging digital governance frameworks, and Chile's AI policy development represent regional momentum that the Dialogue can amplify and connect. The OAS and its specialized agencies provide existing multilateral infrastructure for regional AI governance cooperation that is underutilized. The region's demographic diversity, linguistic richness, and experience navigating complex multilateral environments positions it as a valuable contributor to global norm-setting if given genuine structural access to do so. The Dialogue represents a rare opportunity to ensure that LAC perspectives shape global AI governance rather than simply receive it.

The AI Dialogue occupies a unique position in the international AI governance landscape: it is the only truly universal, multilateral forum explicitly mandated to include governments and all relevant stakeholders on equal footing. That distinction is its most important asset and its most significant responsibility. The Dialogue can advance international cooperation in three specific ways. First, it can serve as a norm-translation mechanism taking the technical and scientific findings of the Independent International Scientific Panel and translating them into policy language accessible to governments at varying levels of technical capacity. This bridge function is currently missing from the governance landscape and is essential for ensuring that evidence informs policy. Second, it can create structured space for regulatory dialogue between jurisdictions that are currently developing AI frameworks in parallel rather than in coordination. The Dialogue does not need to produce a single global AI framework that is neither feasible nor desirable. But it can facilitate the development of interoperability principles that reduce the compliance burden on smaller economies and prevent the emergence of incompatible governance regimes. Third, it can serve as an accountability mechanism for existing commitments. The Global Digital Compact, regional AI strategies, and national AI policies contain significant governance commitments that are rarely tracked or reviewed in a multilateral setting. The Dialogue can create a regular review function that builds shared understanding of implementation progress and gaps. For these functions to be meaningful, the Dialogue must resist the tendency toward lowest common denominator consensus. It should be designed to surface genuine disagreements, not paper over them because durable international cooperation on AI governance requires honest engagement with the political and economic interests that shape national approaches.

Several existing initiatives provide foundations the AI Dialogue should explicitly build upon rather than duplicate. The Organization of American States and its Inter-American Committee against Terrorism (CICTE) have developed regional cybersecurity frameworks and capacity-building programs that are directly relevant to AI governance in the LAC region. The OAS provides multilateral infrastructure, regional legitimacy, and established relationships with member state governments that the Dialogue should leverage for LAC specific engagement and implementation. The OECD AI Principles and the accompanying OECD. AI Policy Observatory represent the most developed existing multilateral framework for AI governance norm setting. The Dialogue should build on this foundation while explicitly expanding its reach to non-OECD countries particularly in LAC, Africa, and Southeast Asia where the Observatory's data coverage and policy engagement remain limited. UNESCO's Recommendation on the Ethics of AI, adopted in 2021 with 193 member states, provides universal normative grounding that the Dialogue should treat as a baseline rather than to relitigate. UNESCO's ongoing implementation support work, particularly in developing countries, is a practical resource the Dialogue Secretariat should integrate. The Global Digital Compact's commitments on AI provide a direct mandate the Dialogue should track and operationalize. Rather than generating new commitments in isolation, the Dialogue can add value by creating a structured review mechanism for GDC implementation. Finally, regional AI governance initiatives Brazil's AI Act process, the LAC Digital Agenda, and emerging frameworks in Colombia and Chile represent genuine regional momentum that the Dialogue should connect and amplify rather than treat as subordinate to global norm setting. The Dialogue's added value lies not in replacing these initiatives but in creating the connective tissue between them facilitating interoperability, reducing duplication, and ensuring that regional and national efforts inform global norms.

Meaningful stakeholder contribution to the AI Dialogue requires moving beyond the standard model of government-led plenaries with civil society observers. Each stakeholder category brings distinct value that the Dialogue's format should be designed to activate. Governments bring legal authority, implementation capacity, and accountability to affected populations. Their role should include not only articulating national positions but sharing implementation experiences what has worked, what has failed, and where international cooperation would accelerate progress. Civil society organizations, particularly those working directly with affected communities, bring documentation of real-world AI harms and the accountability perspective that is often absent from technical governance discussions. They should have substantive roles in thematic sessions not just designated speaking slots in plenary with adequate preparation support and accessible participation modalities. The private sector brings technical expertise and implementation capacity, but also significant conflicts of interest. Their contributions are most valuable when structured around specific technical questions interoperability standards, safety benchmarks, audit methodologies rather than general policy positions. Academia and the technical community bridge scientific evidence and policy application. Their most valuable contribution is translating the Scientific Panel's findings into actionable recommendations accessible to policymakers at varying levels of technical capacity. International and regional organizations including the OAS, African Union, ASEAN, and others provide regional legitimacy and implementation infrastructure that global governance mechanisms lack. They should be treated as structural partners, not peripheral participants. For all stakeholders, meaningful contribution requires adequate lead time, accessible language, and transparent processes for how inputs are incorporated. The Co-Chairs' summary should explicitly reflect the range of perspectives received including dissenting views rather than defaulting to consensus language that obscures genuine disagreement.

Several communities are systematically underrepresented in global AI governance discussions, and their absence is not incidental it reflects the same power asymmetries that AI governance is supposed to address. Indigenous communities worldwide are subject to AI systems that extract, misrepresent, or erase their languages, knowledge systems, and cultural practices yet are almost entirely absent from governance forums. Their inclusion requires more than translation; it requires recognition of Indigenous data sovereignty principles and governance frameworks developed by Indigenous communities themselves. Small island developing states and least developed countries face acute AI governance challenges including climate-related AI applications, limited regulatory capacity, and exposure to AI systems designed for entirely different contexts but lack the technical staff and diplomatic bandwidth to participate meaningfully in multiple simultaneous governance processes. Dedicated capacity support and consolidated engagement pathways are essential. Women and girls, particularly in the Global South, face documented harms from AI systems in areas ranging from facial recognition to content moderation to economic opportunity platforms. Gender-disaggregated analysis of AI impacts should be a baseline requirement, and women-led civil society organizations should have structural roles in the Dialogue. Migrant and displaced communities are among the most exposed to high-stakes AI deployments in border management, asylum processing, and surveillance yet have no formal representation in governance processes. Diaspora organizations and refugee-led civil society groups should be explicitly included. Workers affected by AI-driven automation, particularly in the informal economies that characterize much of the Global South, represent another critical gap. Their perspectives on the socioeconomic implications of AI are essential and currently absent. Inclusion of these communities requires proactive outreach, financial support for participation, accessible formats, and genuine structural roles not tokenism representation in designated slots.

The standard UN conference format prepared statements delivered from a lectern to a half-empty room is poorly suited to the kind of dynamic, evidence-based exchange the AI Dialogue needs to generate. Several format innovations would significantly improve the quality of engagement. Structured deliberation sessions, modeled on deliberative dialogue methodologies, would allow small mixed groups of government representatives, civil society, technical experts, and affected community members to work through specific governance questions together rather than talking past each other in sequential statements. These sessions generate genuine exchange and often surface practical solutions that plenary formats cannot. Problem-based working groups organized around specific governance challenges algorithmic accountability in public services, regulatory interoperability for smaller economies, multilingual AI standards would allow participants with relevant expertise to produce concrete outputs rather than general declarations. These groups should include practitioners from affected regions, not only technical experts from major jurisdictions. Pre-Dialogue regional consultations, beyond the current virtual stakeholder sessions, would allow regional perspectives to be developed collectively and presented as substantive inputs rather than individual statements. The OAS, African Union, and ASEAN frameworks provide ready infrastructure for this. A structured response mechanism where the Co-Chairs or thematic leads respond directly to written submissions and consultation inputs would signal that stakeholder contributions are genuinely informing the Dialogue's agenda rather than being received and filed. Even brief thematic summaries of input patterns would significantly increase trust in the process. Finally, simultaneous interpretation and multilingual written submission options are not optional they are prerequisites for genuine inclusion. A global AI governance forum that operates exclusively in English is structurally excluding most the world's affected populations before the conversation begins.

Several policies, practices, and platforms offer concrete models the AI Dialogue should examine and build upon. Brazil's Lei de Inteligência Artificial, currently advancing through its legislative process, represents the most substantive AI governance framework developed in the Global South. Its approach to risk classification, human rights impact assessment, and accountability mechanisms reflects LAC-specific governance realities and provides a model for other regional governments developing their own frameworks. The European Union's AI Act provides the most comprehensive binding regulatory framework currently in force, with relevance to the Dialogue's work on risk-based governance, prohibited use cases, and conformity assessment. Its extraterritorial implications for LAC economies make it especially important for the Dialogue to address both as a model and as a compliance burden requiring mitigation. The OECD AI Policy Observatory offers the most developed multilingual platform for tracking national AI policies, strategies, and regulatory developments. Expanding its geographic coverage and data quality in LAC, Africa, and Southeast Asia would significantly strengthen the evidence base for the Dialogue. The OAS/CICTE cybersecurity capacity-building program demonstrates a working model for multilateral technical assistance in the LAC region that could be adapted and expanded for AI governance capacity-building leveraging existing institutional relationships, regional legitimacy, and member state engagement infrastructure. At the practice level, algorithmic impact assessments as piloted in Canada's Directive on Automated Decision-Making and emerging frameworks in several LAC jurisdictions provide a practical accountability tool that governments at varying levels of technical capacity can implement. The Dialogue should promote their adoption and develop guidance for implementation in resource-constrained contexts. Finally, civil society documentation initiatives such as Algorithm Watch, AI Now Institute, and regional equivalents provide essential accountability infrastructure that the Dialogue should recognize, support, and draw upon as evidence sources.