H21 Vision

A successful first Global Dialogue on AI Governance should move beyond principles and produce actionable foundations for implementation. First, it should advance interoperable governance approaches across jurisdictions, reducing fragmentation while preserving innovation and respecting national sovereignty. Second, it should establish a baseline of globally recognized good practices for safe, secure, and trustworthy AI—particularly in high-impact sectors such as finance, critical infrastructure, and public services. Third, the Dialogue should address the need for governance mechanisms that operate in real time. AI risks are dynamic, and oversight cannot rely solely on static audits or retrospective compliance reports. Mechanisms capable of detecting anomalies, integrity failures, and emerging systemic risks as they arise should be prioritized. Fourth, the Dialogue should explore how algorithmic risk can be linked to capital allocation and market integrity. Private financial institutions, insurers, investors, and multilateral institutions could play a critical role by integrating algorithmic governance standards into lending, investment, insurance pricing, and institutional ratings. Finally, success would mean creating a roadmap for continued collaboration among governments, industry, academia, and civil society to ensure AI governance evolves as rapidly as the technologies it seeks to oversee.

• Interoperability of governance approaches
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight

My priorities reflect the need to move AI governance from principle-based discussions toward operational, verifiable, and globally interoperable implementation. Safe, secure, and trustworthy AI is foundational, particularly as AI systems increasingly influence financial markets, critical infrastructure, and public services. Interoperability of governance approaches is essential to avoid fragmentation across jurisdictions, which could increase compliance complexity, create regulatory arbitrage, and slow responsible innovation. Transparency, accountability, and human oversight remain central, but these concepts must evolve beyond disclosure and static reporting toward continuous monitoring, real-time risk detection, and mechanisms that enable institutions to verify compliance and system integrity. Finally, the social, economic, ethical, cultural, linguistic, and technical implications of AI are deeply interconnected. AI is not only a technological issue; it affects labor markets, institutional trust, cultural representation, and access to opportunity. Together, these priorities support a governance model that protects citizens and markets while enabling innovation to scale responsibly.

Yes. One important emerging issue is the relationship between algorithmic risk and market integrity. As AI systems increasingly influence investment decisions, lending, insurance pricing, and capital allocation, algorithmic failures or opaque decision-making can create systemic financial and economic risks. A second emerging issue is the need for real-time governance mechanisms. Current governance models often rely on retrospective audits or static compliance reports, which may be insufficient for dynamic AI systems. Future frameworks should consider continuous oversight, anomaly detection, and real-time integrity monitoring. A third issue is the growing importance of cryptographic verification and technical assurance mechanisms, including tools that allow institutions to prove compliance, resilience, and fairness without exposing sensitive intellectual property or data. Finally, AI governance is increasingly linked to digital sovereignty and geopolitical resilience, particularly where AI intersects with critical infrastructure, cybersecurity, and strategic autonomy.

In both Europe and Latin America, as well as across the financial and critical infrastructure sectors, AI governance gaps are creating both significant risks and major opportunities. The most significant challenge is the speed mismatch between technological deployment and regulatory adaptation. AI systems are already influencing capital allocation, financial decision-making, customer interactions, and operational resilience, while governance frameworks often remain fragmented, static, or unclear. In Europe, important advances such as the European Union AI Act are creating a stronger regulatory baseline. However, interoperability across jurisdictions remains a challenge, particularly for multinational companies operating across different legal and compliance regimes. In Latin America, the opportunity is to avoid fragmented or reactive approaches and instead adopt interoperable frameworks aligned with international standards while preserving regional priorities and digital sovereignty. In the financial sector, opaque or poorly governed AI can create market integrity risks, bias in lending or insurance decisions, operational vulnerabilities, and reputational damage. In critical infrastructure, governance failures can affect resilience and public trust. At the same time, these challenges create opportunities for institutions and countries that move early. Organizations that implement trustworthy, transparent, and verifiable AI governance systems can strengthen resilience, improve investor confidence, reduce compliance risk, and gain competitive advantage. The next opportunity lies in moving from retrospective compliance toward real-time oversight, continuous monitoring, and technical assurance mechanisms that allow institutions to verify integrity and compliance as risks emerge.

The AI Dialogue can play a critical role as a neutral global platform for convergence. Its most important contribution would be to help align fragmented governance approaches across jurisdictions by identifying interoperable principles, technical standards, and practical implementation pathways. Beyond principles, the Dialogue can accelerate the transition from high-level ethical commitments to operational governance mechanisms—such as real-time monitoring, technical assurance systems, and verifiable compliance frameworks. It can also create bridges between governments, industry, academia, civil society, and multilateral institutions to address cross-border risks, including cybersecurity threats, systemic market impacts, and risks to critical infrastructure. A particularly valuable role would be to facilitate the development of globally recognized good practices that connect AI governance to market integrity and capital allocation frameworks. Finally, the Dialogue can support emerging economies by sharing best practices, building capacity, and reducing the risk of regulatory fragmentation, ensuring that AI governance is both inclusive and globally relevant.

The AI Dialogue should build upon and connect with existing international initiatives rather than duplicate them. Relevant foundations include the United Nations system's existing work on digital cooperation and global AI governance; the OECD AI Principles; the G7 Hiroshima AI Process; the European Union AI Act as a regulatory benchmark; the NIST AI Risk Management Framework; and sector-specific work by institutions such as the International Monetary Fund, World Bank, and the Bank for International Settlements in finance and systemic risk. The added value of the AI Dialogue would be to act as a coordination layer—connecting regulatory, technical, financial, and geopolitical perspectives. It can also help translate broad principles into actionable and interoperable governance practices, particularly for high-impact sectors such as finance, critical infrastructure, and public services. In addition, it could catalyze the creation of globally recognized clusters of good practices around real-time oversight, algorithmic risk management, and technical assurance mechanisms.

Different stakeholders should contribute according to their comparative strengths. Governments can provide regulatory perspectives and public policy coordination. Industry can contribute practical implementation experience, technical expertise, and insights into operational challenges. Academia and technical experts can support research, standards development, and independent evaluation. Civil society can help ensure that human rights, inclusion, and societal impacts remain central. Financial institutions and insurers can contribute perspectives on market integrity, capital allocation, and systemic risk. To be effective, the AI Dialogue should combine plenary discussions with smaller technical working groups focused on specific themes such as interoperability, real-time oversight, algorithmic risk, and sector-specific governance challenges. It would also be valuable to create public-private task forces to produce actionable outputs, such as model governance frameworks, technical assurance mechanisms, or clusters of good practices. A successful format should prioritize implementation and continuity, not only discussion.

Several voices remain underrepresented in global AI governance discussions. These include stakeholders from the Global South, small and medium-sized enterprises, critical infrastructure operators, financial market actors, insurance and risk professionals, technical auditors, and communities most affected by automated decision-making. Latin America, Africa, and smaller economies are often underrepresented despite facing significant challenges and opportunities related to AI adoption, digital sovereignty, and regulatory capacity. There is also a need for stronger representation of multidisciplinary perspectives, including cybersecurity experts, governance architects, and professionals working at the intersection of AI, finance, and systemic risk. Inclusion can be strengthened through regional consultations, multilingual participation, financial support for participation, and dedicated seats in thematic working groups. A truly global AI governance framework must reflect diverse economic realities, legal traditions, and cultural contexts.

Innovative engagement formats should move beyond traditional panels and encourage practical collaboration. This could include scenario-based workshops, policy simulation exercises, and sector-specific roundtables focused on real-world governance challenges. Technical demonstration sessions could showcase tools for real-time monitoring, technical assurance, and AI risk management. Interactive working labs could bring together policymakers, technologists, investors, and civil society to co-design practical frameworks or governance prototypes. Digital collaboration platforms could allow stakeholders to contribute continuously before and after in-person meetings, ensuring continuity and broader participation. The most effective formats will combine strategic discussion with practical, implementation-oriented outputs.

Examples of effective AI governance approaches already exist and can serve as building blocks for global cooperation. The OECD AI Principles provide a strong foundation for trustworthy AI. The NIST AI Risk Management Framework offers practical guidance for identifying and managing AI risks. The European Union AI Act is establishing an important regulatory benchmark, particularly for risk-based classification and obligations. In Latin America, initiatives such as fAIr LAC and its 3S approach-Safe, Smart, and Sustainable AI-from the Inter American Development Bank Lab, provide practical regional frameworks to assess and guide startups and innovation ecosystems. In my role as Senior Fellow assessing startups within this initiative, I have seen the value of combining innovation support with governance and impact criteria from early stages. In finance, existing governance models around operational risk, model risk management, cybersecurity, and anti-money laundering provide useful analogies for AI oversight and accountability. Promising practices include algorithm registries, model documentation standards, independent audits, red-teaming, and continuous monitoring systems. Looking ahead, more advanced approaches may include real-time oversight mechanisms capable of detecting anomalies and integrity failures as they emerge, rather than relying solely on retrospective compliance. Technical assurance tools such as cryptographic verification, privacy-preserving proofs, and other mechanisms that allow institutions to verify compliance, fairness, and resilience without exposing sensitive intellectual property may also play an important role. Finally, market-based approaches could complement regulation-for example, integrating algorithmic governance standards into investment decisions, lending criteria, insurance pricing, and institutional ratings.