Verizon UK Limited

In my view, the first Global Dialogue on AI Governance would be a success if it delivers a clear, practical foundation for continued international cooperation rather than only broad principles. A strong outcome would be agreement on shared priorities such as AI safety, human rights, capacity-building, and reducing the digital divide, alongside a credible roadmap for follow-up action. It would also be a success if the Dialogue helps countries move from discussion to implementation. That could include common baseline expectations for transparency, risk assessment, and responsible public-sector use of AI, as well as support for countries that need help with skills, infrastructure, and institutional readiness. Just as importantly, success should mean that the process is genuinely inclusive and actionable. If governments, the scientific panel, civil society, and the private sector leave with a shared sense of direction, clearer coordination, and measurable next steps for the 2027 review cycle, that would be a meaningful achievement

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Protection and promotion of human rights

I believe all thematic areas are important and none should be left out. Nonetheless, I selected these four thematic areas because they capture both the immediate risks and the long-term enabling conditions for responsible AI. They also provide a balanced framework that links safety, capability, impact, and coordination. Safe, secure and trustworthy AI: This is the foundation of any credible AI governance effort. Without safety, security, transparency, and accountability, public trust will remain limited and adoption will be uneven. AI capacity-building: This is essential for ensuring that all countries can participate meaningfully in AI development and governance. Capacity-building should include skills, institutions, infrastructure, and access to practical tools. Social, economic, ethical, cultural, linguistic and technical implications of AI: AI affects people differently across contexts, so governance must reflect real-world consequences. This area is important because it helps ensure AI is inclusive, context-aware, and aligned with human values. Interoperability of governance approaches: AI is global, while regulation is often national or regional. Interoperability is therefore critical to avoid fragmentation, reduce compliance burdens, and support coherent international cooperation and widens economic benefits. Together, these four areas reflect my view that AI governance should be safe, inclusive, practical, and globally connected. They also support urgent action while leaving room for diverse national circumstances and priorities.

In my view, there are a few cross-cutting issues that deserve explicit attention. First and most important is AI literacy and public awareness as a cross-cutting issue. Governance will only be effective if policymakers, institutions, and users understand the capabilities and limits of AI. UNESCO AI Competency Framework can serve as a useful reference point for that effort. A governance process will be more effective if people understand not only what AI can do, but also its limits, risks, and appropriate uses. In addition, vendor and third-party accountability is becoming increasingly important, since many AI systems are built or deployed through complex supply chains. Another major one is data governance, including data quality, privacy, provenance, and cross-border data flows, because these affect every AI use case and shape trust, safety, and accountability. Another emerging issue is the governance of frontier and agentic AI, where greater autonomy increases questions around oversight, testing, and responsibility. Finally, I believe environmental impact and the measurement of real-world outcomes are becoming more visible concerns. AI governance should not only ask whether systems are compliant, but also whether they are genuinely beneficial, inclusive, and sustainable. These issues do not sit outside the listed themes; rather, they cut across them and strengthen their practical implementation.

In the UK and EU, the main governance gap is the difference between a more flexible, sector-led UK approach and the EU's more prescriptive, horizontal AI framework. This creates uncertainty for organisations operating across both markets, because compliance expectations, oversight models, and enforcement pathways are not fully aligned. The biggest challenge is fragmentation. In the UK, the absence of a single binding AI law can leave gaps between regulators, while the EU's AI Act provides greater legal clarity but can be demanding to implement, especially for smaller firms and public bodies. A further challenge is that both systems still need stronger practical guidance on issues such as transparency, testing, accountability, and the governance of advanced and generative AI. At the same time, there are important opportunities. The EU's framework can strengthen trust, standard-setting, and market access for compliant AI services, while the UK's more adaptive model can support innovation and faster sector-specific responses. Together, the two approaches can also encourage better interoperability if they converge on shared baseline principles and stronger cooperation between regulators. For the region, the most significant opportunity is to turn regulatory diversity into practical coordination, so that safety, capacity-building, and public trust advance together rather than in separate track.

The AI Dialogue can serve as a practical multilateral space where countries align on shared principles, compare governance approaches, and reduce fragmentation. It can help build common understanding around safe, secure and trustworthy AI, capacity-building, and interoperability, while also ensuring that social, economic, ethical, cultural, linguistic, and technical concerns are reflected in the global conversation. Its greatest value is likely to be in turning dialogue into cooperation. By bringing together governments, industry, and civil society, it can encourage shared baseline expectations, identify governance gaps, and promote coordination with existing regional and international efforts rather than duplicating them. The Dialogue can also help bridge the gap between advanced and developing economies by giving all Member States a seat at the table and by highlighting capacity gaps, digital inclusion, and access to trusted AI tools. That makes it especially important for ensuring that governance is inclusive, balanced, and responsive to different national contexts. In my view, success will depend on whether the Dialogue produces practical follow-up: clearer areas of convergence, usable guidance for policymakers, stronger cooperation on standards, and a credible pathway toward the 2027 review process.

The AI Dialogue should build on existing efforts that already address AI governance, capacity-building, standards, and public engagement. These include multistakeholder forums such as the Partnership on AI, cross-border policy dialogues, UNESCO's AI competency work, and regional regulatory approaches in the UK, EU, and beyond. Its added value would be to provide a universal UN platform where these efforts can be connected, compared, and scaled. That would help reduce fragmentation, identify common baseline principles, and support interoperability across different governance models. It could also bring stronger political legitimacy, wider geographic inclusion, and a clearer link between technical discussion and Member State decision making. The AI Dialogue can also add value by focusing on underrepresented priorities, such as AI capacity gaps in the Global South, AI literacy, and practical pathways for implementation. In that way, it would complement existing initiatives rather than duplicate them, while helping translate good ideas into shared international action.

Different stakeholders can contribute in complementary ways. Governments can set priorities and share policy experience; industry can bring technical expertise, implementation lessons, and resources; academia and the scientific community can provide evidence, benchmarking, and independent analysis; civil society can raise human rights, inclusion, and accountability concerns; and international organisations can help connect the Dialogue to existing initiatives and support follow-up. This multi-stakeholder approach is consistent with the UN's own design for the Dialogue, which envisages thematic discussions with governments and relevant stakeholders, as well as broader plenary exchange. I would recommend a format that is practical, inclusive, and action-oriented. The Dialogue should combine brief high-level remarks with moderated thematic breakout sessions, followed by a plenary session focused on convergence, lessons learned, and next steps. Co-chairing thematic sessions with a Member State and a relevant stakeholder would help balance political ownership with technical and societal expertise. In terms of structure, I would suggest clear pre-read materials, focused questions for each thematic cluster, and a short output note summarising areas of agreement, open issues, and follow-up actions. The process should also allow for participation from diverse regions and levels of development, including remote engagement where needed, so that the Dialogue reflects global realities rather than only the perspectives of a few major actors. The added value of this format would be that it supports genuine exchange, builds trust, and turns dialogue into practical cooperation.

Underrepresented voices include Indigenous peoples, linguistic minorities, persons with disabilities, youth, older persons, conflict-affected communities, and communities in the Global South. Public voices are also often missing, especially those most likely to be affected by AI in education, work, health, and public services. These groups can be included through structured consultations, regional and thematic roundtables, multilingual participation, accessible formats, and travel or connectivity support. The Dialogue should also involve civil society, local researchers, and community representatives in agenda-setting, not only in later-stage commentary. A practical way forward would be to create a standing inclusive advisory group with balanced regional representation and to publish draft outputs for public comment. That would help ensure that the Dialogue reflects lived experience, not only institutional or technical perspectives. Including these voices is not only a matter of fairness; it also improves policy quality by making governance more realistic, context-aware, and responsive to the people most affected by AI.

A few formats can make the AI Dialogue more dynamic and productive. I would recommend short opening statements followed by moderated breakout groups, so participants can move quickly from general principles to practical discussion. This helps keep the conversation focused while still allowing diverse views to emerge. Other effective formats include lightning rounds for concise interventions, fishbowl discussions to widen participation, and structured Q&A sessions with clear speaking limits. Interactive polling or live digital boards can also help surface patterns in real time and give quieter voices a way to contribute. I would also suggest case-based sessions, where each thematic area is explored through a concrete scenario. That would make the discussion more grounded and action-oriented, especially for topics such as capacity-building, interoperability, and trust. A short synthesis session at the end of each block would be useful to capture points of convergence and open questions. In terms of structure, the Dialogue should be inclusive, time-bound, and outcome-focused. It should balance high-level plenary exchange with smaller thematic sessions, ensure regional and stakeholder diversity, and leave space for written submissions before and after the meeting. That combination would help turn the Dialogue into a living process rather than a one-off event.

Examples that could be referenced include national AI governance policies, OECD guidelines, cross-functional oversight committees, sector-specific use-case profiles, and standard-based management systems that turn principles into operational controls. In practice, these approaches help move from high-level commitments to concrete risk assessment, accountability, monitoring, and review. For standards, the most relevant reference is ISO/IEC 42001:2023, which provides a management-system approach for AI governance, including leadership commitment, risk management, impact assessment, lifecycle controls, and supplier oversight. Depending on the organisation's needs, supporting ISO standards such as ISO/IEC 23894 on AI risk management and ISO 31000 on general risk management can also strengthen implementation. The NIST AI Risk Management Framework is another useful model because it is practical, flexible, and designed to help organisations map, measure, manage, and govern AI risks across the full lifecycle. It works well alongside policy templates, internal model registries, human review procedures, testing protocols, and incident response processes. Concrete solutions also include AI governance policies for public bodies, transparency and explainability requirements, bias testing, monitoring dashboards, and defined escalation paths for high-risk use cases. Together, these help ensure that AI is not only innovative, but also safe, accountable, and aligned with public interest.