Pontifical University of Salamanca

It should deliver concrete, inclusive, and actionable outcomes that strengthen global cooperation while respecting diverse national contexts. First, it should establish a shared baseline of principles for AI governance, grounded in international law and aligned with existing frameworks within the United Nations system and the UNESCO Recommendation on the Ethics of Artificial Intelligence. These principles should reinforce commitments to human rights, transparency, accountability, and meaningful human oversight, while allowing for regulatory diversity and innovation. Second, success would mean making tangible progress in bridging the AI divide. The Dialogue should generate clear commitments on capacity-building, including support for developing countries in areas such as data governance, computing infrastructure, and digital skills. This also includes facilitating access to AI resources, open-source tools, and collaborative international mechanisms. Third, the Dialogue should lead to practical cooperation mechanisms rather than purely declarative outcomes. This may include the creation of multi-stakeholder working groups, a roadmap for interoperability across governance approaches, and platforms to share best practices, risk assessment methodologies, and audit standards. Fourth, it should ensure meaningful multi-stakeholder participation, particularly from underrepresented regions, the Global South, and marginalized communities. Inclusivity should be reflected in agenda-setting and decision-making processes. Finally, success would be reflected in clear follow-up and accountability mechanisms, including a timeline for implementation, measurable indicators of progress, and a commitment to continuity. The Dialogue should serve as the foundation for an ongoing, adaptive, and cooperative global AI governance process.

• AI capacity-building
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

The selected priorities reflect a rights-based and inclusive approach to AI governance, with a strong focus on both immediate risks and long-term structural challenges. They are closely aligned with my professional work as an advisor to ParlAmericas and the European Union, as Co-Director of a Master's programme in Ethical Governance of Artificial Intelligence, and as a member of the Women4EthicalAI initiative, where we have developed a policy framework to assess algorithms from a gender perspective. AI capacity-building is essential to address global asymmetries and ensure that all countries, particularly in the Global South, can meaningfully participate in the development, deployment, and governance of AI systems. Without investment in skills, infrastructure, and institutional capabilities, existing inequalities risk being deepened. The social, economic, ethical, cultural, linguistic, and technical implications of AI are equally critical. AI systems are not neutral; they shape and are shaped by societal contexts. Addressing these dimensions is necessary to prevent bias, exclusion, and cultural homogenization, while also ensuring that AI contributes to inclusive economic development. The protection and promotion of human rights is a foundational pillar. AI systems increasingly affect access to essential services, employment, and democratic participation. Ensuring alignment with international human rights law is essential, as reflected in frameworks developed within the United Nations system, including those led by UNESCO. Finally, transparency, accountability, and human oversight are key to operationalizing these principles. They enable auditability, mitigate risks such as algorithmic bias, and ensure that responsibility remains clearly attributable in AI-driven decision-making.

Yes. While the proposed themes are comprehensive, several cross-cutting and emerging issues merit greater visibility. First, the environmental and resource footprint of AI should be explicitly addressed. The development and deployment of large-scale models requires significant energy, water, and raw materials, raising sustainability concerns and reinforcing global inequalities linked to infrastructure and resource extraction. Second, the geopolitical and economic dimensions of AI governance deserve more attention. Issues such as concentration of market power, dependency on a limited number of technology providers, and the risk of regulatory fragmentation affect countries' ability to exercise digital sovereignty and shape their own development pathways. Third, the integrity of information ecosystems is an urgent and evolving challenge. AI systems are increasingly used to generate and amplify disinformation, including foreign information manipulation and interference (FIMI), with direct implications for democratic processes and public trust. In my work with the European External Action Service, specifically under the Foreign Policy Instruments Service, we developed three reports on FIMI in Latin America and the Caribbean. These findings highlighted that the region is highly permeable and vulnerable to foreign interference, underscoring the need for coordinated international responses and stronger resilience mechanisms. Fourth, the impact of AI on labor markets should be considered more holistically. Beyond job displacement, there is a need to address job transformation, the quality of work, and often overlooked forms of labor such as data annotation and content moderation, which raise concerns around fairness, visibility, and working conditions. Fifth, the governance of advanced and emerging technologies, including the intersection between AI and quantum computing, may introduce new security, cryptographic, and risk management challenges that require anticipatory governance approaches. Finally, stronger emphasis should be placed on measurement and evaluation tools, including algorithmic audits and impact assessments, to ensure that governance principles are effectively translated into practice.

From my professional experience working across Europe and Latin America and the Caribbean, governance gaps and uneven AI developments are affecting both regions in distinct but interconnected ways. In Europe, significant regulatory progress—particularly through the EU AI Act—has positioned the region as a global standard-setter. This creates important opportunities for trustworthy AI, legal certainty, and the protection of fundamental rights. However, challenges remain in implementation, including ensuring regulatory coherence, supporting SMEs in compliance, and translating high-level principles such as transparency and accountability into operational practices, including audits and oversight mechanisms. In Latin America and the Caribbean, the primary challenge lies in structural asymmetries. Limited access to infrastructure, data, and technical expertise constrains the region's ability to develop and govern AI systems autonomously. These gaps are compounded by vulnerabilities in institutional capacity and regulatory frameworks. As a result, countries in the region often act as adopters rather than shapers of AI technologies, increasing dependency on external providers and limiting their ability to align AI systems with local priorities and values. At the same time, these challenges create opportunities. There is growing momentum for regional cooperation, capacity-building initiatives, and the adoption of rights-based governance approaches inspired by international frameworks such as those developed within the United Nations system and UNESCO. Additionally, the region has the potential to leapfrog by adopting context-sensitive, inclusive AI strategies, particularly in sectors such as public services, agriculture, and digital governance. Across both regions, bridging governance gaps requires stronger international cooperation, interoperability between regulatory approaches, and sustained investment in human capital and institutional capacity.

The AI Dialogue can play a pivotal role as a neutral, inclusive, and action-oriented platform to strengthen international cooperation on AI governance. First, it can serve as a space to build convergence around core governance principles. By facilitating dialogue among governments and stakeholders with diverse regulatory traditions, the Dialogue can help identify common ground on issues such as human rights protection, transparency, accountability, and safety, while respecting different national contexts. In this sense, it can reinforce and connect existing efforts within the United Nations system, including the work led by UNESCO. Second, the Dialogue can promote interoperability across governance frameworks. Rather than aiming for uniform regulation, it can support the alignment of approaches, standards, and methodologies, enabling countries to cooperate more effectively while reducing fragmentation and regulatory uncertainty. This is particularly important in a landscape where different models, such as those emerging from the European Union and other regions, are evolving in parallel. Third, it can act as a catalyst for capacity-building and resource-sharing. By identifying concrete needs and matching them with existing initiatives, the Dialogue can facilitate technical assistance, knowledge transfer, and access to infrastructure, particularly for developing countries. This is key to ensuring that all countries can meaningfully participate in AI governance. Fourth, the Dialogue can strengthen multi-stakeholder cooperation by bringing together governments, academia, the private sector, and civil society in a structured and continuous process. This inclusivity is essential to address the complex and cross-cutting nature of AI. Finally, it can provide continuity and accountability by establishing follow-up mechanisms, monitoring progress, and fostering iterative exchanges. In doing so, the AI Dialogue can evolve into a long-term platform that supports adaptive, coordinated, and globally legitimate AI governance.

The AI Dialogue should build on and connect with existing international, regional, and multi-stakeholder initiatives that are already advancing AI governance, while providing added value through coordination, inclusivity, and global legitimacy. At the multilateral level, key frameworks developed within the United Nations system, particularly the work of UNESCO on the Recommendation on the Ethics of Artificial Intelligence, offer a normative foundation grounded in human rights. Similarly, regional regulatory efforts such as those led by the European Union provide valuable experience in operationalizing governance through risk-based approaches. At the academic and capacity-building level, initiatives such as the Master's Programme in Ethical Governance of Artificial Intelligence (Universidad Pontificia de Salamanca) bring together a global community of experts, practitioners, and decision-makers, both among faculty and students, fostering interdisciplinary and practice-oriented knowledge exchange. Likewise, the IALAB UBA (Universidad de Buenos Aires) plays an important role in advancing research, training, and policy dialogue on AI governance in Latin America. In addition, multi-stakeholder networks, expert groups, and regional cooperation mechanisms—including those focused on digital policy, innovation, and development—are critical spaces that the Dialogue should actively connect with to avoid duplication and leverage existing expertise. The added value of the AI Dialogue lies in its universal and inclusive nature. Unlike other initiatives, it can provide a single global platform where all countries—regardless of their level of technological development—can participate on equal footing. It can act as a bridge between fragmented efforts, facilitate interoperability across frameworks, and ensure that capacity-building, knowledge-sharing, and policy coordination are scaled globally. By connecting these initiatives, the Dialogue can move from principles to practice, strengthening coherence and accelerating collective action in AI governance.

Different stakeholders can contribute to the AI Dialogue by bringing complementary forms of expertise, practical experience, and regional perspectives. Governments can provide policy direction and regulatory insight; the private sector can contribute technical knowledge and implementation experience; academia can offer independent research and critical analysis; and civil society can ensure that societal impacts and human rights considerations remain central. From the perspective of academic and capacity-building initiatives, the experience of the Master's Programme in Ethical Governance of Artificial Intelligence (now in its fourth edition at Universidad Pontificia de Salamanca) offers relevant lessons for the structure and format of the Dialogue. The programme brings together a diverse, global community of experts and practitioners from the public sector, private sector, international organizations, and academia. This diversity has proven essential to foster meaningful, interdisciplinary exchanges grounded in real-world challenges. One key recommendation is to adopt a hybrid and modular structure that combines high-level plenary discussions with smaller, thematic working groups. In the Master's programme, this format has enabled both strategic reflection and in-depth, solution-oriented dialogue. Interactive methodologies—such as case studies, policy labs, and problem-solving sessions—should be integrated to move beyond declarative statements toward actionable outcomes. Another important element is continuity. Rather than a one-off event, the Dialogue should be structured as an iterative process with regular sessions, follow-up mechanisms, and opportunities for participants to contribute between meetings. This mirrors the sustained engagement model of the programme, where learning and collaboration evolve over time. Finally, the Dialogue should prioritize inclusivity and accessibility, ensuring participation from diverse regions and backgrounds. Leveraging digital tools and hybrid formats can help broaden engagement and reduce barriers to participation, while maintaining high-quality, structured exchanges.

Several voices and perspectives remain underrepresented in global debates on AI governance, which limits the inclusiveness and legitimacy of resulting frameworks. First, stakeholders from the Global South—particularly from Latin America and the Caribbean, Africa, and parts of Asia—are often underrepresented in agenda-setting and standard-setting processes. This is not only a matter of participation, but of influence. Their inclusion requires sustained investment in capacity-building, funding for participation in international forums, and mechanisms that enable them to shape—not just respond to—global governance discussions. Second, marginalized and vulnerable communities, including women, indigenous peoples, and linguistic minorities, are frequently excluded. This is especially critical given that AI systems can reproduce and amplify existing inequalities. Initiatives such as Women4EthicalAI demonstrate the importance of integrating gender perspectives into AI governance, including through tools such as gender-sensitive algorithmic audits. Expanding similar approaches to other dimensions of inequality is essential. Third, workers involved in the AI value chain—such as data annotators and content moderators—are rarely included in policy discussions, despite being directly impacted by AI systems and their governance. Their perspectives are key to addressing issues related to labor conditions, fairness, and the sustainability of AI systems. Fourth, local governments and public sector practitioners are often overlooked, even though they are responsible for implementing AI systems in areas such as public services, education, and healthcare. Their operational insights are critical for designing realistic and effective governance frameworks. To address these gaps, the AI Dialogue should adopt proactive inclusion strategies: targeted outreach, financial and technical support for participation, multilingual engagement, and structured mechanisms—such as dedicated panels or advisory groups—that ensure these voices are not only present but influential in decision-making processes.

Innovative participation formats are essential to move the AI Dialogue from passive exchange toward meaningful, solution-oriented engagement. First, policy labs and co-creation workshops can foster dynamic interaction among stakeholders. These formats bring together diverse participants to work on concrete challenges—such as designing accountability mechanisms or capacity-building strategies—and produce actionable outputs. This approach has proven effective in academic and executive education settings, including interdisciplinary programmes that combine legal, technical, and policy perspectives. Second, scenario-based simulations and foresight exercises can help participants explore the implications of emerging technologies and governance choices. By engaging stakeholders in structured "what-if" scenarios—such as the deployment of high-risk AI systems or cross-border regulatory conflicts—these formats encourage anticipatory governance and collaborative problem-solving. Third, interactive digital platforms can enhance inclusivity and real-time engagement. Tools that enable live polling, moderated Q&A, and collaborative drafting allow participants from different regions to contribute actively, even in hybrid or fully online settings. This is particularly important to ensure participation from underrepresented regions. Fourth, multi-stakeholder working groups with rotating leadership can sustain engagement beyond plenary sessions. These groups can focus on specific thematic areas and produce iterative outputs, such as guidelines, toolkits, or policy recommendations, feeding into the broader Dialogue process. Fifth, peer-learning formats—such as case study exchanges between countries or institutions—can facilitate the sharing of practical experiences, including successes and failures in AI governance implementation. Finally, incorporating evaluation and feedback loops into these formats is key. Continuous assessment of participation quality and outcomes can help refine the Dialogue over time, ensuring it remains responsive, inclusive, and impactful.

Several existing policies, practices, and initiatives provide valuable examples of effective and actionable AI governance. The EU AI Act represents a comprehensive, risk-based regulatory framework that classifies AI systems according to their potential impact and establishes corresponding obligations. Its emphasis on high-risk systems, transparency, and accountability offers a concrete model for operationalizing trustworthy AI while fostering innovation. At the global level, the UNESCO Recommendation on the Ethics of Artificial Intelligence provides a normative, human rights-based framework adopted by a broad range of countries. It is complemented by tools such as ethical impact assessments, which support practical implementation and monitoring. At the local level, New York City Local Law 144 on automated employment decision tools introduces mandatory bias audits, offering a concrete example of how algorithmic accountability can be enforced in practice. This type of approach is particularly relevant for addressing discrimination and ensuring fairness in AI systems. In addition, policy-oriented research and global policy dialogue are generating practical solutions. For example, a policy brief presented in the context of the G20 Brazil 2024 highlighted algorithmic auditing as a key mechanism to prevent and mitigate bias in AI systems. This work emphasized the need for standardized audit methodologies, independent oversight, and the integration of gender and diversity perspectives into evaluation processes. From a capacity-building and multi-stakeholder perspective, academic and applied initiatives also play a key role. The Master's Programme in Ethical Governance of Artificial Intelligence (Universidad Pontificia de Salamanca) brings together global experts and practitioners to bridge theory and practice, fostering the development of governance skills across sectors. Similarly, the IALAB UBA (Universidad de Buenos Aires) promotes interdisciplinary research, training, and policy dialogue on AI governance in Latin America. In addition, emerging practices such as algorithmic audits, regulatory sandboxes, and multi-stakeholder policy labs are proving effective in testing, evaluating, and refining governance approaches in real-world contexts. Together, these examples illustrate that effective AI governance requires a combination of regulatory frameworks, practical tools, and sustained investment in knowledge, capacity, and collaboration.