Independent Expert (Banking & Financial Services – Risk Management) – personal capacity

A successful first Global Dialogue on AI Governance would be defined by tangible progress from principles to implementation. This includes agreement on a minimum, globally interoperable baseline for AI governance, such as risk-based classification of AI systems linked to proportionate safeguards (e.g., human oversight, transparency, and monitoring). Success would also entail practical guidance to support adoption across jurisdictions, particularly through capacity-building mechanisms, shared tools, and knowledge transfer to help bridge AI divides. Ensuring that governance approaches are inclusive, multi-stakeholder, and aligned with human rights principles will be critical to building trust. In addition, the Dialogue should establish ongoing collaboration mechanisms, including pathways for continued technical input, cross-border coordination, and iterative refinement of governance practices as AI evolves. Ultimately, success lies in enabling countries at all levels of readiness to implement responsible, scalable, and interoperable AI governance frameworks.

• Safe, secure and trustworthy AI
• AI capacity-building
• Transparency, accountability, and human oversight

These priorities reflect a practical need to move from high-level AI principles to consistent and scalable implementation across jurisdictions. From a practitioner perspective in the banking and financial services sector, this requires governance approaches that are both risk-based and adaptable across diverse environments. Ensuring safe, secure and trustworthy AI is foundational, particularly as AI systems increasingly influence critical decision-making across sectors. Transparency, accountability, and human oversight are essential to building trust and ensuring that AI systems operate in alignment with ethical standards and human rights. These elements enable effective governance, auditability, and risk mitigation, especially in high-impact use cases. The interoperability of governance approaches is critical to avoid fragmentation across regulatory regimes. A degree of global consistency-such as common risk classification frameworks and proportionate control expectations can support cross-border adoption while allowing flexibility for national contexts. Finally, AI capacity-building is essential to ensure that all countries, regardless of their level of technological maturity, can participate meaningfully in AI development and governance. This includes access to tools, skills, and knowledge-sharing mechanisms to bridge existing gaps. Together, these priorities support the development of risk-based, proportionate, and globally adaptable AI governance frameworks, enabling both responsible innovation and effective risk management across diverse environments.

While the listed themes comprehensively address key dimensions of AI governance, several cross-cutting and emerging issues warrant further attention. First, there remains a critical need to translate high-level principles into operational and implementable governance frameworks. Many global discussions emphasize ethical commitments, but less focus is placed on how these can be consistently applied across organizations, sectors, and jurisdictions in a practical and scalable manner. Second, the importance of adaptive and forward-looking governance is increasing as AI technologies evolve rapidly. Governance approaches should enable continuous monitoring, iterative improvement, and responsiveness to emerging risks and use cases. Third, greater emphasis is needed on global interoperability of governance and assurance practices, including alignment in standards, validation approaches, and reporting mechanisms. This can help reduce fragmentation, build trust, and support cross-border collaboration. Finally, there is a need to further integrate societal and human-centered considerations, including fairness, inclusion, and equitable access, into governance frameworks to ensure that AI benefits are broadly shared. Addressing these cross-cutting issues can strengthen the effectiveness, inclusiveness, and long-term sustainability of AI governance globally.

Governance gaps and uneven developments in AI are creating both challenges and opportunities across the banking and financial services sector. A key challenge is the lack of consistent and interoperable governance frameworks across jurisdictions, which leads to fragmentation in regulatory expectations. Organizations operating across regions face increased complexity in aligning AI practices with differing standards, creating inefficiencies and potential gaps in oversight. Another challenge is the difficulty of translating high-level principles into operational practices. While expectations around transparency, accountability, and human oversight are well established, implementing these consistently across use cases remains complex, particularly as AI systems evolve. At the same time, these gaps present opportunities. There is growing momentum to develop risk-based, proportionate governance approaches that enable organizations to prioritize resources based on the potential impact of AI systems. This supports more effective risk management while allowing innovation to continue. In addition, increased focus on interoperability and capacity-building creates an opportunity to establish more consistent global practices, reduce duplication, and enhance trust across stakeholders. Overall, addressing these governance gaps can enable the sector to balance innovation with responsible oversight, supporting resilience, trust, and sustainable adoption of AI technologies.

The AI Dialogue can serve as a neutral, inclusive platform to align global efforts and translate high-level principles into practical, interoperable governance approaches. By bringing together governments, industry, academia, and civil society, it can help identify common priorities and develop shared understanding of risks, safeguards, and implementation challenges. A key role of the Dialogue is to facilitate convergence across jurisdictions, reducing fragmentation by encouraging alignment on core elements such as risk-based approaches, transparency expectations, and human oversight. This can support more consistent governance while allowing flexibility for national contexts. The Dialogue can also advance cooperation by promoting capacity-building and knowledge-sharing, enabling countries at different levels of AI readiness to participate meaningfully in governance and adoption. In addition, it can support ongoing, adaptive collaboration, providing a mechanism for continuous engagement as AI technologies evolve. Ultimately, the AI Dialogue can help move from principles to coordinated action, strengthening trust, enabling resp

The AI Dialogue can build upon existing international and multi-stakeholder efforts, including work by the OECD on AI principles, UNESCO on AI ethics, the G20 initiatives on digital economy and AI, and regional regulatory developments such as the European Union's AI governance frameworks. It can also connect with technical standard-setting bodies and industry-led initiatives that contribute to AI safety, testing, and best practices. The added value of the AI Dialogue lies in its universality and inclusiveness. Unlike existing initiatives that may be regional or sector-specific, it provides a global platform where all countries can participate equally. It can act as a coordinating layer, bridging policy, technical, and operational perspectives, and helping translate diverse initiatives into coherent and interoperable approaches. By fostering alignment, reducing duplication, and promoting shared tools and guidance, the Dialogue can strengthen global cooperation and accelerate the practical implementation of responsible AI governance.

Different stakeholders can contribute by bringing complementary perspectives and expertise to the AI Dialogue. Governments can provide policy direction and regulatory insight; the private sector can share practical implementation experience; academia and the technical community can contribute research and methodological rigor; and civil society can ensure human rights, inclusion, and societal impacts remain central. To be effective, the AI Dialogue should adopt a multi-layered structure. This could include: -Plenary sessions for high-level alignment and agenda-setting -Thematic working groups focused on priority areas such as safety, interoperability, and capacity-building -Technical roundtables to explore implementation challenges and emerging risks -Regional consultations to reflect diverse contexts and needs Outputs should be action-oriented, including summaries, practical guidance, and voluntary frameworks that support implementation. Establishing mechanisms for ongoing engagement and follow-up will be important to ensure continuity and adaptability as AI evolves.

Despite growing global engagement, several voices remain underrepresented in AI governance discussions. These include stakeholders from developing countries, particularly those with limited technical and institutional capacity, as well as small and medium-sized enterprises (SMEs), which often lack the resources to participate in global forums. In addition, frontline practitioners and implementers, those responsible for deploying and managing AI systems in real-world environments, are often underrepresented compared to policy and research communities. Their insights are critical for understanding operational challenges and ensuring that governance frameworks are practical and implementable. Greater inclusion can be achieved through targeted outreach, capacity-building initiatives, and accessible participation formats, including virtual engagement. Providing financial and technical support, translation services, and simplified materials can help lower barriers to participation. Ensuring that diverse perspectives are systematically incorporated into outcomes will strengthen the legitimacy, relevance, and effectiveness of global AI governance.

To foster meaningful and dynamic engagement, the AI Dialogue should incorporate interactive and participatory formats that complement traditional plenary discussions. These could include: -Scenario-based thematic workshops to explore real-world governance challenges and policy trade-offs -Multi-stakeholder roundtables to facilitate dialogue across governments, industry, academia, and civil society -Case-based discussions highlighting cross-sector implementation experiences and practical lessons learned In addition, hybrid formats combining in-person and virtual participation can broaden access and inclusivity. Enabling virtual participation from contributors across regions, including those who may not be able to attend in person, can help ensure more diverse perspectives are represented. Digital collaboration tools can further support continuous engagement before and after formal sessions, allowing contributors to provide inputs, share feedback, and remain engaged over time. These formats can complement high-level discussions by enabling more focused, implementation-oriented engagement. They also support collaboration across disciplines and help translate broad principles into practical, actionable outcomes that are adaptable to different national and sectoral contexts.

Effective AI governance is supported by a combination of global principles and practical frameworks that translate high-level commitments into implementation. At the global level, frameworks such as the OECD AI Principles and the UNESCO Recommendation on the Ethics of AI establish widely recognized foundations for trustworthy, human-centric AI, emphasizing transparency, accountability, and respect for human rights. Complementing these, frameworks such as the National Institute of Standards and Technology AI Risk Management Framework provide more practical and structured guidance on managing AI-related risks, including governance, measurement, and continuous improvement. In practice, some organizations are beginning to adopt risk-based governance approaches, including classification of AI use cases based on potential impact and linking them to proportionate controls such as human oversight, documentation, and ongoing monitoring. These approaches can support more consistent and scalable implementation across sectors.