Independent Professional

Success would require moving beyond declaratory commitments toward concrete governance architecture decisions that is both inclusive and enforceable. Three outcomes would be decisive. First, the Dialogue should produce a shared framework that recognizes AI governance as inseparable from human rights obligations — not as a technical annex to existing instruments, but as a foundational reorientation of how states and institutions design AI decisional systems. This means operationalizing rights-based standards within algorithmic architectures, not only in ex-post regulatory review. Second, success requires genuine structural inclusion of the Global South — particularly the LAC, African, and Southeast Asian regions — not as recipients of norms designed elsewhere, but as co-architects of governance frameworks. This demands dedicated capacity-building commitments with time-bound benchmarks and adequate financing mechanisms, anchored to existing development cooperation frameworks. Third, the Dialogue must produce a clear institutional mandate for follow-up. Without a persistent multilateral body with monitoring functions, the outcomes risk repeating the fragmentation that has characterized AI governance to date. A success scenario is one where this Dialogue initiates — not concludes — a sustained global process.

• AI capacity-building
• Protection and promotion of human rights
• Interoperability of governance approaches

The three priorities selected reflect a governance logic organized around effectiveness, inclusion access and sustainabikity. Human rights protection establishes the normative foundation: governance without binding rights obligations produces accountability gaps that compound existing inequalities. Interoperability of governance approaches addresses the most immediate institutional failure - regulatory fragmentation creates arbitrage conditions that favor the most resourced actors and exclude most of the Global South from meaningful norm-setting. AI capacity-building, however, must be reframed. The dominant discourse treats it as training and institutional strengthening - necessary but insufficient. The deeper deficit is infrastructural: access to compute, semiconductors, energy systems, and foundational models. Without addressing this layer, capacity-building reproduces dependency rather than resolving it. The Global Dialogue should advance an international pact that recognizes AI infrastructure as a primary access good for governance and development - not a market commodity distributed by geopolitical leverage.

Two structural issues remain largely absent from the thematic framework of GA Resolution 79/325. The first is AI infrastructure as a global public good. Current governance debates focus on regulation, safety, and ethics - but systematically avoid the upstream question of who controls the physical and digital infrastructure on which AI runs: compute capacity, semiconductor supply chains, energy systems, and access to foundational models. This concentration - in a handful of private corporations and two or three jurisdictions - is the foundational governance problem. Without an international pact that treats AI infrastructure access as a primary condition for democratic participation and sovereign governance and protects about its impact and at the same time garantying sustentability, all downstream regulatory frameworks will reproduce the dependency relationships they claim to address. The analogy to previous multilateral moments is instructive: as with telecommunications, energy access, and internet governance, the failure to establish equitable infrastructure access at an early stage generates path dependencies that become structurally irreversible. The second is decisional sovereignty - the capacity of states, communities, and individuals to meaningfully shape the AI systems that govern their lives. This is distinct from data sovereignty or regulatory autonomy. It refers to the structural conditions under which agency over AI architecture can actually be exercised. Without compute access, without the ability to train or audit models, without energy infrastructure - decisional sovereignty is nominal. The Dialogue must treat infrastructure access and decisional sovereignty as prerequisites, not afterthoughts, of any legitimate global AI governance framework.

The Latin America and the Caribbean region illustrates with particular acuity how governance gaps in the three selected areas interact and compound one another. On interoperability, LAC faces a structural dilemma: the region is simultaneously subject to the extraterritorial effects of the EU AI Act, influenced by US standards through trade and investment relationships, and internally fragmented across national regulatory initiatives of uneven technical capacity. This multi-layered exposure without coordination mechanisms produces regulatory uncertainty that disproportionately affects smaller economies and limits the region's collective negotiating power in global norm-setting processes. On human rights, the deployment of AI systems in public administration — predictive policing, social benefit allocation, migration management — is advancing faster than rights-based oversight frameworks. The region has a documented history of algorithmic systems replicating and amplifying structural discrimination along racial, gender, and socioeconomic lines, often through technologies procured from external providers without transparency requirements or redress mechanisms. On AI infrastructure, the challenge is most structurally acute. LAC accounts for a marginal share of global compute capacity, semiconductor production, and foundational model development. Energy infrastructure constraints further limit the region's ability to host or operate competitive AI systems domestically. This creates a dependency on infrastructure controlled by actors with no accountability to LAC populations or governments — effectively outsourcing the material conditions of AI governance to external private interests. The opportunity lies precisely in the region's position: LAC has the institutional tradition — through CELAC, ECLAC, and bilateral EU-LAC frameworks — to articulate a collective governance position that bridges the Global South and multilateral institutions. A coordinated LAC voice in this Dialogue, grounded in infrastructure equity and rights-based governance, could shift the terms of the global conversation.

The AI Dialogue can play a role that no existing forum currently fulfills: establishing a legitimate, universal, and politically binding process for AI governance coordination. Existing mechanisms — the OECD AI Principles, the Bletchley process, the EU AI Act's extraterritorial reach — are either club-based, technically oriented, or geographically limited. The Dialogue's value lies precisely in its UN anchoring, which gives it the normative legitimacy to produce frameworks applicable across jurisdictions with radically different regulatory capacities. Concretely, the Dialogue should advance three cooperation functions. First, a coordination mechanism for regulatory interoperability — not harmonization, but mutual recognition frameworks that prevent regulatory arbitrage. Second, a multilateral infrastructure compact that treats compute access, energy, and foundational model availability as development cooperation priorities, financed through existing ODA mechanisms and new dedicated instruments. Third, a monitoring architecture with genuine Global South participation — not as observers but as co-designers of evaluation frameworks. The Dialogue must resist becoming a high-level declaration process. Its added value is institutional: creating the preconditions for sustained, accountable, and inclusive AI governance cooperation.

Several existing frameworks offer institutional foundations the Dialogue should deliberately connect with rather than duplicate. The UNESCO Recommendation on the Ethics of AI (2021) provides the broadest normative consensus to date and should serve as the rights-based floor for any Dialogue outcomes. The ITU's AI for Good platform offers infrastructure for technical exchange that could be expanded toward governance functions. The OECD AI Policy Observatory provides comparative regulatory data that should be made accessible and actionable for non-OECD members. The EU-LAC Digital Alliance represents an emerging model of North-South digital cooperation that the Dialogue could scale. The Dialogue's added value relative to these initiatives is political weight and universality. What it can provide that none of these can is a General Assembly-anchored process with binding follow-up potential. It should therefore position itself as the coordination apex of existing initiatives — not a parallel track — while ensuring that institutions designed by and for high-income countries do not set the default terms of a nominally global framework.

Meaningful participation in the AI Dialogue requires moving beyond the standard multi-stakeholder model, which in practice privileges well-resourced actors — large technology firms, Northern civil society organizations, and states with established AI policy infrastructure. The format should be structured around differentiated participation tracks with explicit equity safeguards. State delegations should include technical and human rights experts, not only diplomatic representatives. Civil society participation must be resourced — travel, translation, and preparation support — particularly for organizations from the Global South. Indigenous communities, whose data, languages, and cultural heritage are disproportionately implicated in AI training datasets, require dedicated representation mechanisms, not token inclusion. Structurally, the Dialogue should adopt a working group architecture organized by thematic area, with rotating co-chair arrangements that guarantee regional balance. Outputs from each working group should be subject to public comment periods before adoption, ensuring that affected communities — not only organized stakeholders — can respond. The process design itself is a governance question. A Dialogue that replicates exclusionary participation patterns in its own architecture will produce frameworks that legitimate rather than redress existing power asymmetries.

The most structurally underrepresented voices in global AI governance are not simply absent — they are systematically excluded by the design of participation itself. Small island developing states and least developed countries lack the diplomatic and technical capacity to engage meaningfully across the multiplying AI governance forums. Indigenous communities face the compounded exclusion of having their cultural and linguistic heritage incorporated into AI systems without consent, while being absent from the spaces where governance of those systems is decided. Women and gender-diverse communities — particularly in the Global South — face differentiated AI harms in labor markets, public administration, and digital safety, yet remain underrepresented in both technical and policy governance spaces. Inclusion requires structural intervention, not good intentions. Concretely: dedicated funding for participation from underrepresented regions and communities; translation and interpretation in all six UN languages plus regional languages; pre-Dialogue capacity-building on AI governance concepts for delegations without specialized staff; and formal quotas for civil society and affected community representation in working group structures. Inclusion is not a procedural courtesy — it is a governance quality requirement. Frameworks designed without the participation of those most affected by AI systems will be both less legitimate and less effective.

The AI Dialogue risks replicating the format failures of previous multilateral processes: panel-heavy plenaries, pre-negotiated texts, and participation asymmetries that render formal openness substantively empty. Three format innovations would meaningfully change the quality of engagement. First, deliberative working sessions structured around specific governance problems — not thematic overviews — where participants from different stakeholder categories work toward concrete output texts. Problem-centered design produces more actionable outcomes than thematic review. Second, structured South-South exchange tracks where states and civil society from the Global South develop shared positions prior to and during the Dialogue. This shifts the dynamic from Global South actors responding to Northern-framed agendas to articulating autonomous governance proposals with collective weight. Third, real-time public participation mechanisms — not symbolic consultations — where affected communities can submit evidence, challenge claims, and respond to draft outputs during the Dialogue process. AI governance is consequential enough to warrant genuine democratic accountability in its own design. The format is not a logistical question. It is a political one: who gets to shape the norms, under what conditions, and with what resources to do so meaningfully.

Several policy approaches demonstrate what rights-based, structurally inclusive AI governance can look like in practice. The EU AI Act establishes the most comprehensive risk-based regulatory framework to date, with its prohibited uses and high-risk classification system offering a replicable model - though its extraterritorial effects and compliance burden on smaller jurisdictions require explicit mitigation in global adoption. Brazil's AI Bill and LGPD illustrate how a Global South country can develop rights-grounded AI regulation with genuine democratic process, including extensive public consultation. It offers a model for governance that is neither a copy of EU frameworks nor ungoverned. The CADE-ANATEL-ANPD coordination in Brazil demonstrates institutional interoperability at the national level - relevant for the Dialogue's interoperability agenda. The African Union's AI Continental Strategy represents an important effort to articulate collective AI governance priorities from a regional perspective rather than adapting external frameworks. Finally, the emerging EU-LAC Digital Alliance provides a nascent model for North-South digital cooperation that could be deepened into an infrastructure access framework - directly addressing the compute and energy deficit that underlies the LAC region's governance gap. The common thread across effective approaches is institutional intentionality: governance that treats rights, equity, and accountability as design requirements, not afterthoughts.