Mediglobal Costa Rica

A successful first Global Dialogue on AI Governance should deliver concrete, implementable, and globally interoperable outcomes, rather than high-level principles alone. First, it should establish a baseline global framework for safe, secure, and trustworthy AI, with harmonized minimum standards for transparency, auditability, and risk classification—especially for high-impact sectors such as healthcare. This includes clear requirements for explainability, human oversight, and accountability across borders. Second, the dialogue must advance data governance models that recognize individuals as active digital citizens, ensuring ownership, consent, and controlled sharing of personal data. Interoperability frameworks—similar to emerging health data spaces—should be prioritized to enable secure cross-border data exchange while preserving sovereignty and privacy. Third, a key outcome should be the creation of certification and credentialing pathways for AI-enabled professionals and systems, enabling trusted international collaboration. This is particularly relevant in fields like digital health, where qualified professionals may operate across jurisdictions under shared standards. Fourth, the dialogue should produce mechanisms for real-world implementation, including regulatory sandboxes, pilot programs, and public-private partnerships that accelerate responsible innovation while maintaining safeguards. Finally, success requires a commitment to equitable access and capacity building, ensuring that low- and middle-income countries are not left behind in AI adoption, governance, and infrastructure. In summary, the dialogue should move from principles to actionable governance, enabling a globally aligned, human-centered, and innovation-driven AI ecosystem.

• Interoperability of governance approaches
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models
• Protection and promotion of human rights

The selected priorities reflect the need to build a globally coherent, human-centered, and operational AI governance framework, particularly in high-impact sectors such as healthcare. Interoperability of governance approaches is essential to avoid fragmentation between jurisdictions. Without aligned standards, cross-border collaboration-especially in areas like digital health, data exchange, and AI-assisted care-becomes inefficient or unsafe. Interoperability enables trusted global ecosystems. Protection and promotion of human rights must remain the foundation of AI governance. AI systems should reinforce dignity, privacy, and autonomy, ensuring that individuals maintain control over their data and are not reduced to passive data subjects, but recognized as active digital participants. Transparency, accountability, and human oversight are critical to building trust. High-risk AI systems must be explainable, auditable, and subject to clear responsibility frameworks. Human-in-the-loop models are particularly important in clinical and decision-making environments where outcomes directly affect lives. Finally, open-source software, open data, and open AI models can accelerate innovation, democratize access, and reduce global inequities. When properly governed, openness fosters collaboration, reproducibility, and capacity-building, especially for low- and middle-income regions. Together, these priorities move AI governance from abstract principles to practical, scalable, and inclusive implementation, ensuring that technological advancement remains aligned with human values and global equity.

Yes. Several critical cross-cutting issues remain underrepresented and should be explicitly addressed to ensure effective AI governance. First, data ownership and patient/citizen sovereignty require stronger recognition. Current frameworks often emphasize protection, but not true control. Individuals should be empowered as active decision-makers over their data, including dynamic consent, traceability of access, and the ability to authorize cross-border use-particularly relevant in sectors like healthcare. Second, cross-border accountability and jurisdiction in AI-enabled services remains insufficiently defined. As AI systems and professionals increasingly operate across jurisdictions, clear legal frameworks are needed to determine responsibility, liability, and regulatory oversight in transnational contexts. Third, clinical-grade validation and real-world performance monitoring of AI systems is essential, especially in health. Beyond initial approval, continuous auditing, post-deployment surveillance, and outcome-based evaluation should be standardized globally to ensure safety and effectiveness. Fourth, human-AI collaboration models deserve more attention. Governance should not only focus on restricting AI, but on defining optimal integration between human expertise and AI systems, particularly in high-stakes decision-making environments. Fifth, digital infrastructure asymmetry is a major emerging issue. Without addressing disparities in data infrastructure, connectivity, and technical capacity, global AI governance risks widening inequalities rather than reducing them. Addressing these cross-cutting dimensions would strengthen AI governance by moving toward a more operational, equitable, and globally interoperable system, capable of responding to real-world implementation challenges.

In my country and region, as well as in the healthcare sector, AI governance gaps are creating both significant risks and transformative opportunities. One of the main challenges is the lack of interoperable governance frameworks. Existing regulations are often fragmented, sector-specific, and not designed for cross-border data flows or AI-enabled services. This limits collaboration and slows down innovation, particularly in digital health, where continuity of care increasingly depends on secure data exchange. Another critical gap is limited operationalization of transparency and accountability. While principles exist, there are insufficient mechanisms for real-world auditing, traceability of AI decisions, and assignment of liability. In clinical environments, this creates uncertainty and reduces trust in AI adoption. Additionally, human rights protections are not yet fully adapted to AI contexts, especially regarding data ownership, dynamic consent, and patient control over health data. Individuals are still largely treated as passive data sources rather than active participants in decision-making. There is also a capacity and infrastructure gap, including limited technical expertise, uneven digital infrastructure, and insufficient access to validated AI tools. This risks widening inequalities between regions and health systems. However, these gaps also present key opportunities. Countries like Costa Rica can position themselves as innovation hubs for responsible AI in healthcare, adopting interoperable, patient-centered data governance models aligned with international frameworks. There is strong potential to develop cross-border digital health ecosystems, enabling secure collaboration between regions such as Latin America and Europe. By addressing these challenges through coordinated policy, investment in infrastructure, and international alignment, AI governance can evolve into a driver of equitable, efficient, and high-quality healthcare systems.

The AI Dialogue can play a pivotal role as a global coordination platform that moves from principles to operational alignment. First, it can act as a bridge between regulatory ecosystems, facilitating convergence across regions with different legal and institutional frameworks. By promoting interoperability of standards—particularly in high-impact sectors like healthcare—it can enable safe cross-border deployment of AI systems and services. Second, the Dialogue can serve as a neutral convening space for governments, academia, industry, and civil society to co-develop practical governance tools, including model regulations, audit frameworks, and certification pathways. This reduces duplication and accelerates global readiness. Third, it can promote trusted data-sharing ecosystems, supporting the development of interoperable data spaces that respect privacy, sovereignty, and human rights while enabling innovation and research collaboration. Fourth, the Dialogue can strengthen capacity-building and knowledge transfer, particularly for low- and middle-income countries, ensuring that global AI governance is inclusive and not dominated by a few regions. Finally, it can function as a mechanism for continuous monitoring and adaptive governance, identifying emerging risks, sharing best practices, and updating standards in response to technological evolution. In this sense, the AI Dialogue should not only facilitate discussion, but actively enable a coordinated, scalable, and globally trusted AI governance architecture.

The AI Dialogue should build upon and connect with existing global and regional initiatives that are already shaping AI governance. Key examples include the OECD AI Principles, UNESCO Recommendation on the Ethics of AI, the EU AI Act, and emerging frameworks such as the European Health Data Space (EHDS). Additionally, global health and digital initiatives led by organizations like WHO and World Bank provide important foundations for sector-specific governance. Regional and national efforts in digital health, open data ecosystems, and AI regulatory sandboxes should also be integrated to ensure practical implementation. The added value of the AI Dialogue lies in its ability to connect these fragmented initiatives into a coherent, interoperable global system. Rather than creating new isolated frameworks, it can align standards, promote mutual recognition mechanisms, and facilitate cross-border certification of AI systems and professionals. Furthermore, it can introduce real-world implementation pathways, such as pilot programs, cross-regional testbeds, and public-private collaboration models, ensuring that governance translates into practice. Importantly, the Dialogue can amplify the voice of underrepresented regions by promoting equitable participation and shared governance, helping to reduce global asymmetries in AI development and deployment. Ultimately, its value is in transforming existing efforts into a coordinated, inclusive, and action-oriented global AI governance ecosystem.

A successful first Global Dialogue on AI Governance should deliver concrete, implementable, and globally interoperable outcomes, rather than high-level principles alone. First, it should establish a baseline global framework for safe, secure, and trustworthy AI, with harmonized minimum standards for transparency, auditability, and risk classification—especially for high-impact sectors such as healthcare. This includes clear requirements for explainability, human oversight, and accountability across borders. Second, the dialogue must advance data governance models that recognize individuals as active digital citizens, ensuring ownership, consent, and controlled sharing of personal data. Interoperability frameworks—similar to emerging health data spaces—should be prioritized to enable secure cross-border data exchange while preserving sovereignty and privacy. Third, a key outcome should be the creation of certification and credentialing pathways for AI-enabled professionals and systems, enabling trusted international collaboration. This is particularly relevant in fields like digital health, where qualified professionals may operate across jurisdictions under shared standards. Fourth, the dialogue should produce mechanisms for real-world implementation, including regulatory sandboxes, pilot programs, and public-private partnerships that accelerate responsible innovation while maintaining safeguards. Finally, success requires a commitment to equitable access and capacity building, ensuring that low- and middle-income countries are not left behind in AI adoption, governance, and infrastructure. In summary, the dialogue should move from principles to actionable governance, enabling a globally aligned, human-centered, and innovation-driven AI ecosystem.

Several critical voices remain underrepresented in global AI governance discussions, limiting both legitimacy and real-world effectiveness. First, low- and middle-income countries (LMICs)—particularly from Latin America, Africa, and parts of Asia—are often underrepresented in decision-making spaces, despite being directly impacted by AI deployment and regulatory gaps. Their perspectives are essential to avoid governance models that are not globally applicable. Second, frontline professionals (e.g., clinicians, educators, public sector operators) are rarely included, even though they are the ones who implement and interact with AI systems in high-stakes environments. Their experiential knowledge is crucial for designing practical and safe governance frameworks. Third, patients, citizens, and end-users remain largely excluded as active participants. Current approaches tend to treat individuals as data subjects rather than as active digital stakeholders with agency over how their data is used and shared. Fourth, small and medium-sized enterprises (SMEs) and local innovators face structural barriers to participation compared to large technology actors, limiting diversity in innovation and governance perspectives. To address this, inclusion must shift from symbolic to structural participation. This includes funded participation mechanisms, regional representation hubs, multilingual engagement strategies, and digital platforms that enable continuous, asynchronous contributions. Additionally, governance processes should incorporate user-centered and consent-driven models, ensuring that individuals can actively shape AI systems and data ecosystems. By integrating these voices, AI governance can become more equitable, context-aware, and operationally effective, reflecting the diversity of real-world environments in which AI is deployed.

Innovative engagement formats should prioritize real-world application, continuous collaboration, and outcome-driven participation, moving beyond traditional discussion-based models. First, policy sandboxes and regulatory testbeds are essential. These environments allow stakeholders to pilot AI systems and governance frameworks under controlled conditions, generating evidence-based insights and accelerating safe implementation—particularly in high-impact sectors such as healthcare. Second, multi-stakeholder simulation labs can be used to model complex scenarios (e.g., cross-border data use, AI clinical decision-making, or liability cases). These exercises enable policymakers, technologists, and practitioners to collaboratively test responses to real-world challenges. Third, the Dialogue should leverage persistent digital collaboration platforms that allow asynchronous participation, transparent documentation, and iterative co-creation of policies. This ensures that engagement is continuous and inclusive, rather than limited to periodic events. Fourth, sector-specific innovation labs should be established to bring together frontline professionals, engineers, regulators, and end-users to co-design solutions tailored to real operational contexts. Additionally, challenge-driven formats—such as global calls for solutions, hackathons, and innovation challenges aligned with governance priorities—can mobilize diverse talent and generate scalable, practical outputs. Finally, all engagement formats should be structured to produce tangible deliverables, including validated frameworks, interoperable standards, pilot results, and policy-ready recommendations. By integrating these approaches, the AI Dialogue can evolve into a dynamic, implementation-oriented platform, capable of translating global discussions into actionable and measurable impact.

Several existing policies and approaches provide strong foundations for effective AI governance, particularly when they combine principles with operational mechanisms. The EU AI Act offers a risk-based regulatory model that classifies AI systems according to their potential impact, enabling proportionate oversight. Complementing this, the OECD AI Principles and the UNESCO Recommendation on the Ethics of AI provide globally recognized normative frameworks centered on human rights, transparency, and accountability. In the health sector, emerging initiatives such as the European Health Data Space (EHDS) demonstrate how interoperable data governance can enable secure, cross-border data sharing while preserving privacy and individual control. This model is particularly relevant for scaling AI in clinical and research environments. Another effective approach is the use of regulatory sandboxes, implemented in various jurisdictions, which allow safe testing of AI systems under supervision. These environments help bridge the gap between innovation and regulation by generating real-world evidence. Additionally, open-source ecosystems and collaborative platforms have proven valuable in accelerating innovation and transparency, especially when combined with governance safeguards such as auditability and documentation standards. A key best practice is the integration of continuous monitoring and post-deployment evaluation, ensuring that AI systems remain safe, effective, and aligned with evolving standards over time. The most impactful approaches are those that move beyond static regulation toward adaptive, interoperable, and implementation-oriented governance, combining legal frameworks, technical standards, and real-world validation. Scaling these models globally-while ensuring inclusivity and contextual adaptation-can significantly strengthen AI governance and enable responsible innovation across sectors.