Ilustre y Nacional Colegio de Abogados de México (INCAM)

The success of the first Global Dialogue on AI Governance will depend on its ability to move from normative convergence toward operational architectures that enable implementation across diverse national contexts. First, success should be measured by the establishment of a shared baseline for interoperable AI governance, aligned with UNGA Resolution A/RES/79/325. This requires not only agreement on principles, but also initial alignment on how regulatory frameworks, technical standards, and institutional practices can function coherently across jurisdictions without requiring uniformity. Second, the Dialogue should deliver concrete implementation pathways, particularly for emerging and middle-income economies. This includes pilot frameworks, regulatory sandboxes, and cross-regional cooperation mechanisms that allow governance models to be tested under real conditions. Without this layer, global AI governance risks remaining declarative rather than operational. Third, a successful outcome would include the establishment of a permanent coordination mechanism focused on interoperability, capacity-building, and standards alignment, ensuring continuity and structured cooperation beyond the Dialogue itself. Equally important is the meaningful inclusion of Global South actors as co-designers of implementation models, not only as participants. Countries positioned between regulatory systems and development realities such as Mexico can serve as strategic bridges between frameworks and deployment. Finally, success should be reflected in the definition of priority areas with measurable implementation indicators, particularly in domains such as human rights protection, mitigation of synthetic harm, and accountable public sector use of AI. Ultimately, the Dialogue will be successful if it establishes AI governance not only as a shared vision, but as a functioning, human-centered and interoperable system capable of being implemented, adapted, and scaled globally.

• Interoperability of governance approaches
• AI capacity-building
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

These four priorities represent the core pillars for advancing a functional and implementable global AI governance architecture, as identified by the Artificial Intelligence Commission of the Ilustre y Nacional Colegio de Abogados de México (INCAM). 1. Interoperability of governance approaches: This is the most urgent priority to prevent regulatory fragmentation. Enabling compatibility between legal frameworks, technical standards, and institutional practices (e.g., EU-Mexico-UN) is essential to ensure that AI governance operates coherently across jurisdictions without requiring uniformity. 2. AI capacity-building: Governance frameworks are only as effective as the capacity to implement them. For emerging and middle-income economies, capacity-building is the enabling condition that translates global commitments into enforceable national systems and reduces structural asymmetries in AI governance. 3. Protection and promotion of human rights: In line with UNGA Resolution A/RES/79/325, AI governance must be inherently human-centered. This requires embedding human rights safeguards directly into the AI lifecycle, particularly in areas such as synthetic harm, algorithmic bias, and protection of vulnerable populations. 4. Transparency, accountability, and human oversight: These elements constitute the operational backbone of trustworthy AI. Robust accountability mechanisms and meaningful human oversight are essential to ensure that AI systems remain subject to democratic control and legal responsibility. By focusing on these priorities, we contribute to a governance model that is interoperable, human-centered, and implementation-oriented, capable of translating global standards into effective local practice while remaining scalable across different contexts.

Yes. A critical cross-cutting issue that remains insufficiently addressed is the structural gap between normative frameworks and operational implementation, particularly across jurisdictions with uneven regulatory and institutional capacities. While existing thematic areas capture key dimensions of AI governance, they do not fully address operational interoperability as a systemic function, nor the need for mechanisms that translate global principles into enforceable, context-sensitive practices. This gap is especially visible in emerging and middle-income countries, where governance challenges are not only regulatory, but also institutional and infrastructural. An additional emerging issue is the rise of synthetic harm and authenticity risks, including deepfakes, AI-enabled manipulation, and the erosion of trust in digital information ecosystems. These risks cut across human rights, security, and governance, and require more integrated responses linking standards, verification systems, and legal accountability. Furthermore, it is essential to address asymmetries in governance capacity and influence. Without intentional design, global AI governance may reinforce existing power imbalances, where standards are defined by a limited number of actors without sufficient participation from the Global South. The Dialogue would therefore benefit from incorporating the concept of "implementation architectures": structured governance models that define how systems operate in practice, including compliance layers, institutional coordination, and cross-border alignment, while preserving digital sovereignty. Addressing these issues is essential to ensure that AI governance evolves not only as a normative framework, but as a functioning, interoperable, and scalable global system.

Gaps in AI governance are already shaping both the risks and opportunities faced by countries like Mexico and similar middle-income contexts. A central challenge is the misalignment between rapidly advancing global regulatory frameworks and limited domestic implementation capacity. While international standards particularly from the EU and multilateral processes are evolving quickly, their translation into local legal, institutional, and technical systems remains uneven. This results in regulatory fragmentation, weak enforcement, and reduced trust in AI deployment, especially in high-impact sectors such as public services and digital information ecosystems. At the same time, there are significant capacity constraints related to technical expertise, institutional coordination, and oversight mechanisms. These gaps limit the ability to effectively govern risks such as algorithmic bias, opaque decision-making, and the growing impact of synthetic content. However, these challenges also present a strategic opportunity. Countries positioned between advanced regulatory systems and emerging market realities such as Mexico can serve as operational bridges for interoperability, testing how global standards can be adapted, implemented, and scaled across diverse contexts. This includes the use of pilot frameworks, regulatory sandboxes, and cross-regional cooperation models. Additionally, the increasing centrality of human rights, transparency, and accountability provides an opportunity to embed these principles from the outset into governance systems, rather than retrofitting them later. Addressing these dynamics requires a shift toward coordinated, interoperable, and implementation-oriented governance models, capable of reducing asymmetries while enabling inclusive and responsible AI deployment.

The Global Dialogue on AI Governance can play a pivotal role in transforming international cooperation from fragmented exchanges into a coherent and operational system of coordination. First, the Dialogue can act as a convergence platform for interoperability, enabling alignment between regulatory frameworks, technical standards, and governance practices across jurisdictions. Rather than seeking uniformity, it can facilitate structured compatibility, allowing diverse systems to function together effectively. Second, it can serve as a catalyst for implementation-oriented cooperation, promoting mechanisms such as cross-regional pilot projects, regulatory sandboxes, and shared capacity-building initiatives. These tools are essential to bridge the gap between global commitments and real-world deployment, particularly for emerging and middle-income countries. Third, the Dialogue can strengthen multistakeholder coordination by establishing a continuous space for governments, international organizations, private sector actors, academia, and civil society to co-design governance solutions. This includes ensuring that Global South actors are not only included, but actively engaged in shaping interoperable governance models. Additionally, the Dialogue can contribute to reducing asymmetries in governance capacity and influence, by facilitating access to knowledge, standards, and best practices, while promoting more balanced participation in global decision-making processes. Ultimately, its value lies in enabling a shift from ad hoc cooperation toward a structured, sustained, and implementation-driven model of international collaboration, capable of supporting a human-centered, interoperable, and scalable global AI governance ecosystem.

The AI Dialogue should function as a Global Interoperability Hub, building upon existing pillars to avoid duplication and move from principles to implementation: Science and Implementation: Align with the Global Digital Compact (GDC) mechanisms specifically the International Scientific Panel and the Global Fund for AI to ensure policy is evidence-based and supported by funding for capacity building. Ethics and Diagnostics: Leverage UNESCO's 2021 Recommendation and its Readiness Assessment Methodology (RAM). This provides the Dialogue with immediate, country-level diagnostic data from over 50 nations (including Latin America), identifying specific governance gaps without starting from scratch. Benchmark Interoperability: Use the EU AI Act as a primary regulatory benchmark. The Dialogue must facilitate a "translation layer" between European risk-based standards and the development needs of the Global South, ensuring market access and regulatory compatibility. Regional Best Practices: Incorporate operational templates from Asia (e.g., ASEAN's governance guides and Singapore's deployment toolkits) and Latin America's emerging national strategies. These regions serve as critical "implementation testbeds" for governance under real-world constraints. Technical Deployment: Bridge with the ITU's AI for Good and the IGF to ensure continuity in technical standards and multistakeholder legitimacy. The Added Value of the AI Dialogue: Unlike fragmented regional efforts, the Dialogue provides a permanent multilateral space to create a "Rosetta Stone" for regulatory compatibility. Its unique value lies in: Harmonization: Establishing a shared taxonomy that makes different governance regimes (EU, US, Global South) interoperable. Execution: Converting high-level ethics into measurable roadmaps and "governance-by-design" pilots that reduce the compliance gap for developing nations. Sovereignty + Integration: Ensuring that AI for Development (A4D) remains compatible with global standards while protecting national digital sovereignty

An effective AI Dialogue must move beyond symbolic inclusion and ensure that all stakeholders contribute to decision-making, implementation, and accountability. 1. Structured Stakeholder Contributions Different actors should have clearly defined roles: Governments: Provide political direction, commit to national implementation, and align regulatory frameworks. Private sector: Contribute technical expertise, data, and scalable solutions, while committing to compliance and transparency. Academia and scientific community: Supply independent evidence, risk assessments, and foresight capabilities. Civil society: Ensure legitimacy, represent affected communities, and monitor human rights impacts. Participation should be organized around functional roles, not only representation, to avoid fragmented discussions. 2. Modular and Action-Oriented Structure The Dialogue should be structured in three interconnected layers: Policy Track: High-level political alignment on principles, risk approaches, and interoperability. Technical Track: Translation of principles into standards, tools, and deployable governance mechanisms (linked to platforms such as AI for Good). Implementation Track: Country-level pilots and capacity-building programs, especially in the Global South, using existing diagnostics (e.g., UNESCO RAM). Each track should produce concrete outputs (roadmaps, pilot commitments, technical guidelines) rather than general statements. 3. Continuous and Measurable Process The Dialogue should not be a one-off event, but a permanent mechanism with: Defined milestones and reporting cycles, Measurable indicators of progress, A follow-up system linking commitments to implementation support. 4. Inclusive by Design, Not by Exception Inclusion must be operationalized through: Regional representation (Latin America, Africa, Asia) in agenda-setting, Financial and technical support for participation, Integration of local realities into global standards. Added Value: This structure transforms inclusion into co-creation of interoperable governance, ensuring that diverse actors do not only participate but shape and implement outcomes.

Across global AI governance debates, participation remains skewed toward well-resourced governments, large technology firms, and institutions from a limited set of regions. The following voices and perspectives are consistently underrepresented, despite being essential for legitimacy and implementation: Underrepresented voices / communities Global South implementers: public servants, regulators, and local innovators who must operationalize governance with limited capacity and infrastructure. Communities most exposed to harm: women and girls; children and youth; migrants and refugees; Indigenous peoples; persons with disabilities; LGBTQI+ communities—especially where AI amplifies discrimination, surveillance, or synthetic violence. Workers and labor institutions: trade unions, informal-sector representatives, and frontline workers affected by automation, algorithmic management, and workplace surveillance. Small and medium enterprises (SMEs) and startups: actors who face "compliance exclusion" when rules are designed for large firms only. Local-language and cultural perspectives: communities affected by linguistic bias, content moderation failures, and the erosion of cultural rights in data-driven systems. Security and justice practitioners outside major powers: local law enforcement oversight bodies, victim-support organizations, and judicial actors dealing with evidence integrity, deepfakes, and digital violence. How to include them (in practice, not symbolically) Representation with decision influence: reserve seats in agenda-setting bodies and drafting groups for regional and affected-community representatives, not only speaking slots. Funded participation: establish travel/remote participation grants, translation/interpretation, and accessibility support (including sign language and disability accommodations). Regional pre-dialogues + rotating hubs: convene structured consultations in Latin America, Africa, and Asia that feed directly into the global agenda with published inputs. Participation through outcomes: require every workstream to include an "impact and feasibility" review led by affected communities and implementers, with a formal response obligation. Safe mechanisms for sensitive harms: create protected channels for testimony and evidence-sharing on AI-enabled violence and surveillance, with safeguards for at-risk groups. This shifts inclusion from visibility to co-design and accountability, ensuring governance reflects real-world constraints and harms.

Most proposals focus on inclusion as presence. The missing piece is integration how to connect political dialogue, technical standards, and real-world implementation into a single operating system. The AI Dialogue should differentiate itself by introducing formats that synchronize all layers of governance: 1. The "End-to-End Governance Cycle" (Flagship Format) A structured sequence where a single issue (e.g., synthetic media, AI in health) moves through three consecutive spaces: Political Alignment Room: governments define the regulatory objective and risk threshold. Technical Translation Room: experts convert this into standards, audit methods, and deployable tools (linked to AI for Good). Implementation Room: countries and partners commit to pilots, funding, and timelines. This creates a closed loop from principle to execution, within the same Dialogue. 2. Interoperability Corridors (Cross-Regional Tracks) Instead of isolated discussions, the Dialogue should establish live corridors (e.g., EU–Latin America, Asia–Africa) where actors jointly design compatible governance pathways. These corridors function as real negotiation and alignment spaces, not panels. 3. Global AI Governance Control Room (Real-Time Coordination) A central, continuously updated platform that tracks: national initiatives, regulatory developments, pilot projects and commitments. This allows stakeholders to see the system as a whole, identify gaps, and coordinate action dynamically. 4. "Commitment-to-Delivery" Mechanism Every announced initiative must be linked to: a responsible actor, a timeline, measurable indicators. Progress is reviewed in subsequent sessions, shifting the Dialogue from promises to accountability. Added Value: This model transforms participation into a synchronized governance architecture. The Dialogue becomes not a forum of fragmented voices, but the place where politics, technology, and implementation converge into coordinated global action.

To achieve effective AI governance, we must move from static regulation to Dynamic Interoperability Systems. The following practices represent the "gold standard" for a synchronized global response: 1. Risk-Based Interoperability (The EU AI Act + Global South Context) The EU AI Act is the essential benchmark for risk classification. However, a best practice is not its literal adoption, but the creation of "Regulatory Translation Tables" that allow national frameworks (like those emerging in Mexico and Brazil) to remain compatible with European standards while addressing local development constraints. This prevents market fragmentation. 2. Distributed Implementation via UNESCO's RAM UNESCO's Readiness Assessment Methodology (RAM) is the most effective diagnostic tool for identifying institutional gaps in the Global South. A best practice is to make RAM results investment-ready, linking identified gaps directly to international funding and technical assistance programs. 3. Agile Testing through "Cross-Border Sandboxes" While national Regulatory Sandboxes (Singapore, Spain) are vital, the next frontier is the Multi-jurisdictional Sandbox. This allows startups and governments to test AI systems across different regulatory environments simultaneously, accelerating the creation of global safety standards. 4. Incident Transparency and Collective Learning The AI Incident Database and the OECD's monitoring frameworks must be transformed into a Global Early Warning System. Governance is ineffective if it doesn't learn from failures in real-time. 5. Sovereign Digital Humanism (The "Mexico Case") A robust approach is the integration of ethical principles into legislative architecture, as seen in Mexico's multistakeholder efforts. This treats Humanism not as a philosophical constraint, but as a technical requirement for market trust and social stability. Why this approach works: Effective governance is not about more rules; it is about coordinated execution. By linking risk-based regulation (EU) with diagnostic rigor (UNESCO) and sovereign priorities (LatAm), we create a governance model that is both globally interoperable and locally relevant.