Thames Valley AI (TVAI) Hub and Henley Business School, University of Reading

A successful first Global Dialogue on AI Governance could be measured not by consensus but by the quality of shared understanding, trust, and momentum it creates among stakeholders and across regions. First, success would mean establishing a common baseline of priorities for AI governance that reflects diverse economic, cultural, and institutional contexts. This includes shared recognition of key risks, such as safety, bias, misuse, and concentration of power, alongside the transformative opportunities AI offers for productivity, public services, and societal wellbeing. Importantly, voices from emerging economies, SMEs, academia, and regional innovation ecosystems should be meaningfully reflected rather than merely consulted. Second, the Dialogue should result in practical pathways for collaboration rather than abstract principles. This could include agreement on mechanisms for ongoing dialogue between governments, industry, and universities; alignment between global norms and regional regulatory experimentation; and support for knowledge exchange on implementation, skills, and evaluation. From our perspective at Thames Valley AI Hub and Henley Business School, based in Reading, United Kingdom, connecting global governance discussions to regional business (including SMEs) realities is essential for credibility and impact. Third, success would involve a clear commitment to responsible innovation: governance that enables adoption and competitiveness while embedding accountability, transparency, and human oversight. The Dialogue should help dispel a false dichotomy between regulation and innovation by highlighting evidence-based, proportionate approaches that organisations can realistically adopt. Finally, the Dialogue should conclude with defined next steps: agreed areas for further work, identified stewards for follow-up, and mechanisms for tracking progress. A living, inclusive process, rather than a one-off event, will be critical. If the Dialogue achieves these outcomes, it will have laid a strong foundation for global AI governance that is actionable, inclusive, and anchored in real-world impact.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Transparency, accountability, and human oversight

Our selection reflects the practical role that Thames Valley AI Hub and Henley Business School play in connecting global AI governance aims with organisational practice, skills development, and regional innovation. Safe, secure and trustworthy AI is a priority because trust is a prerequisite for meaningful adoption. Businesses and public sector organisations require clarity on safety expectations, risk management, and assurance mechanisms to deploy AI with confidence. Governance discussions must therefore focus on approaches that are credible, proportionate, and grounded in evidence rather than abstract principles. AI capacity-building is essential to ensure that AI governance is not limited to those with the greatest resources. This includes developing leadership capability, workforce skills, and institutional understanding across regions, sectors, and stages of digital maturity. As a business school and a regional AI hub, we see first-hand that gaps in skills, literacy, and organisational readiness remain a major barrier to responsible AI adoption. Interoperability of governance approaches is critical for organisations operating across borders and regulatory regimes. Fragmentation increases costs, uncertainty, and the risk of uneven standards. We prioritise dialogue that supports alignment between international norms, national regulation, and local experimentation, enabling organisations to innovate responsibly while remaining compliant. Transparency, accountability, and human oversight underpin all of the above. These elements are central to maintaining legitimacy, protecting individuals, and ensuring that AI systems remain aligned with human values and decision-making. In practice, organisations need workable guidance on governance structures, documentation, and oversight models that can be embedded into everyday operations. Together, these priorities reflect our commitment to AI governance that is inclusive, implementable, and capable of supporting both innovation and societal trust.

Yes. While the listed themes capture many critical dimensions of AI governance, there are several cross-cutting and emerging issues that merit explicit attention. First, the organisational and institutional readiness for AI governance cuts across all themes. Many challenges arise not from the absence of principles, but from limited capability within organisations to operationalise them. This includes governance maturity, leadership accountability, internal controls, procurement practices, and the integration of AI oversight into existing risk and compliance structures. Without addressing this, global norms risk remaining aspirational rather than actionable. Second, the role of SMEs and regional innovation ecosystems is insufficiently foregrounded. Smaller organisations are often both the fastest adopters and the least resourced to navigate complex governance expectations. Ensuring that governance frameworks are scalable, proportionate, and supported by regional intermediaries is essential for inclusive and competitive AI adoption. Third, AI evaluation, measurement, and evidence generation are emerging issues. Policymakers and organisations alike lack shared metrics and methodologies for assessing AI risks, benefits, and real-world impacts over time. Stronger alignment on evaluation practices would support trust, learning, and regulatory coherence. Fourth, the interaction between AI governance and sustainability goals deserves greater emphasis. This includes environmental impacts such as energy use and resource consumption, as well as longer-term societal and economic sustainability. AI governance should be aligned with broader sustainable development objectives rather than treated in isolation. Accreditation can also help evidence that AI systems are being assessed against recognised environmental, social and governance criteria. Finally, the pace of technological change relative to governance cycles remains a structural challenge. Mechanisms for adaptive governance, regulatory learning, and continuous dialogue between developers, users, and regulators are increasingly important. Addressing these cross-cutting issues would strengthen the existing thematic areas by improving implementation, inclusivity, and long-term effectiveness of global AI governance efforts.

In the UK, and particularly within the Thames Valley region and the business and higher education sectors we work with, governance gaps in the selected thematic areas create both significant challenges and important opportunities. For safe, secure and trustworthy AI, the main challenge is uneven organisational readiness. While large organisations are increasingly investing in AI assurance and risk management, many SMEs lack the resources and expertise to interpret evolving expectations around safety, security, and risk. This creates adoption hesitancy and uneven standards for some. Others rush AI services to market with little governance awareness or concern. At the same time, there is an opportunity for the UK to lead in developing practical, proportionate approaches to AI assurance that can be embedded across organisations of different sizes. In relation to AI capacity-building, skills shortages remain a critical constraint. There is strong demand for AI literacy among leaders, managers, and policymakers, not just technical specialists. Without this, governance requirements are often misunderstood or treated as a compliance burden rather than a strategic capability. The opportunity lies in universities, business schools, and regional hubs playing a stronger role in translating governance principles into executive education, workforce training, and applied research. For interoperability of governance approaches, organisations operating across markets face growing complexity as regulatory and voluntary frameworks evolve at different speeds. Fragmentation increases costs and legal uncertainty, particularly for scaling businesses. However, this also creates an opportunity for the UK to act as a convenor and testbed for alignment between international norms, national regulation, and sector-led standards. Finally, organisations struggle with questions about documentation, explainability, procurement, and ongoing oversight of deployed AI systems. Overall, these developments highlight the importance of grounding AI governance in real organisational contexts, supported by skills development, regional ecosystems, and continuous dialogue between policy, academia, and practice.

The AI Dialogue can play a pivotal role in advancing international cooperation on AI governance by acting as a trusted, neutral space for sustained engagement across governments, academia, business, and civil society. First, the Dialogue can support cooperation by building shared understanding. Differences in regulatory maturity, economic context, and technological capacity often lead to fragmented approaches. By enabling open exchange on priorities, risks, and lessons learned, the Dialogue can help establish common reference points that respect diversity while reducing misunderstanding and duplication. Second, the AI Dialogue can facilitate alignment and interoperability rather than uniformity. International cooperation does not require identical rules, but it does require coherence. The Dialogue can promote convergence around key concepts such as risk management and accountability, and support mutual learning between jurisdictions at different stages of development. This would be particularly help SMEs. Third, the Dialogue can strengthen cooperation by connecting global norms to implementation realities. Universities, business schools, and regional innovation hubs play an important intermediary role in translating governance objectives into organisational practice. The Dialogue can amplify these perspectives, ensuring that international cooperation is informed by how AI is actually developed, procured, and used within organisations and economies. Fourth, the Dialogue can lower barriers to participation by supporting inclusive capacity-building. Sharing knowledge, tools, and educational resources can help ensure that countries, regions, and organisations with fewer resources are not excluded from shaping or implementing AI governance frameworks. Finally, the Dialogue can advance cooperation by creating continuity. Clear follow-up mechanisms, working groups, and signals of long-term commitment are essential to move beyond one-off discussions toward durable collaboration. In this way, the AI Dialogue can foster international cooperation that is practical, inclusive, and capable of evolving alongside AI technologies themselves.

The AI Dialogue should build upon and connect existing global, regional, and sector based initiatives, while adding value through coordination, inclusivity, and continuity. At the global level, the Dialogue can build on initiatives such as UN agency work on AI and digital cooperation, OECD AI principles and policy observatories, UNESCO's work on AI ethics, and other multilateral and multi stakeholder forums addressing AI risk and opportunity. These efforts have generated important norms and guidance, but they often operate in parallel. The AI Dialogue can add value by acting as a connective layer, improving awareness, coherence, and mutual learning across these initiatives. At the national and regional level, the Dialogue should connect with regulatory and policy innovation taking place in different jurisdictions, including regulatory sandboxes, assurance frameworks, and sector specific governance models. Linking these experiences can surface practical lessons about what works in implementation, particularly for businesses, public sector organisations, and smaller actors navigating complexity. Partnerships involving universities, business schools, innovation hubs, and industry networks are also critical. These actors play a translational role between high level governance objectives and day to day organisational practice through education, skills development, applied research, and SME engagement. The AI Dialogue can amplify these intermediary perspectives, which are often underrepresented in global forums. The added value of the AI Dialogue lies in four areas. First, providing an inclusive and neutral space that brings together policy, practice, and education. Second, focusing on interoperability and implementation rather than duplicating principles. Third, supporting capacity building and shared tools that lower barriers to participation. Fourth, ensuring continuity through follow up mechanisms that sustain collaboration over time. By connecting existing efforts and focusing on practical impact, the AI Dialogue can strengthen global AI governance while respecting diversity of context and approach.

Different stakeholders can contribute to the AI Dialogue by bringing complementary perspectives, evidence, and capabilities, provided the format is designed to encourage meaningful participation rather than symbolic representation. Governments and intergovernmental organisations can contribute policy experience, regulatory insights, and mechanisms for coordination across borders. Their role is critical in articulating public interest objectives, sharing lessons from implementation, and identifying areas where alignment or mutual recognition is possible. Businesses and industry groups, including SMEs, can contribute practical insights into how AI is developed, procured, and deployed in real organisational settings. Their participation helps ensure that governance discussions are grounded in operational realities, innovation incentives, and supply chain considerations, rather than theoretical models alone. Universities, business schools, and research institutions can contribute independent analysis, evidence, and education capacity. They are well placed to bridge policy and practice through applied research, skills development, and evaluation of governance approaches. Civil society and professional bodies can surface societal concerns, ethical considerations, and impacts on rights, inclusion, and trust, helping ensure that governance frameworks remain legitimate and people-centred. To support these contributions, we recommend an AI Dialogue structure that combines: • Plenary sessions to establish shared understanding and highlight strategic priorities • Thematic and cross-sector roundtables focused on implementation challenges, interoperability, and case studies • Regional and sectoral perspectives, ensuring voices beyond major technology centres are included • Working groups or communities of practice tasked with follow-up, knowledge sharing, and practical outputs • Clear mechanisms for continuity, including feedback loops, published summaries, and pathways into future dialogues A balanced, multi-layered structure will help ensure the AI Dialogue is inclusive, action-oriented, and capable of translating international cooperation into real-world impact.

Several important voices and perspectives remain underrepresented in global discussions on AI governance, despite being central to effective and legitimate outcomes. Small and medium sized enterprises and regional innovation ecosystems are often missing from global forums. SMEs are major users and integrators of AI, yet they face disproportionate challenges in interpreting and implementing governance expectations. Inclusion can be improved through dedicated SME-focused sessions, regional case studies, and partnerships with local AI hubs, chambers of commerce, and universities that can aggregate and represent their experiences. Practitioners responsible for implementation, such as managers, procurement leads, risk officers, and public sector delivery teams, are also underrepresented. Governance discussions frequently focus on policy design rather than execution. Including these operational roles would ground the Dialogue in practical realities and support more implementable outcomes. Structured practitioner roundtables and evidence-based case sharing could facilitate this. Education and skills institutions, including business schools, vocational providers, and professional bodies, are not sufficiently visible in AI governance debates. These actors play a critical role in building leadership capability, workforce readiness, and ethical literacy. Their inclusion as delivery partners rather than observers would strengthen capacity-building efforts globally. Communities outside major technology centres, including those in regions with limited resources or differing cultural and linguistic contexts, remain underrepresented. Participation barriers include cost, access, and relevance. The Dialogue could address this through regional engagement mechanisms, hybrid participation models, and collaboration with trusted local intermediaries. AI governance discussions benefit from combining technical, legal, economic, organisational, and behavioural insights. To include these voices, the AI Dialogue should prioritise inclusive design. This includes targeted outreach, support for participation, structured roles for intermediaries, and formats that value applied experience alongside policy expertise. Broadening participation in this way would improve the legitimacy, implementation, and real-world impact of global AI governance.

Innovative engagement formats are essential to ensure that the AI Dialogue moves beyond set-piece statements towards substantive exchange and learning across stakeholders. One effective format would be policy to practice labs, where mixed groups of policymakers, businesses, academics, and civil society work through concrete governance challenges using real or simulated case studies. This would help surface practical tensions, trade-offs, and implementation insights that are often absent from plenary discussions. Curated multi-stakeholder roundtables can also foster deeper engagement. Rather than open panels, these sessions should be deliberately composed across regions, sectors, and roles, with skilled facilitation to encourage peer-to-peer dialogue and mutual learning. Limiting session size and focusing on defined questions can improve the quality of interaction. The Dialogue could benefit from regional and sectoral showcases, where representatives present short, focused examples of governance approaches, regulatory experiments, or organisational practices, followed by structured discussion. This would highlight the diversity of context while enabling comparative learning. Practitioner and leadership clinics are another innovative format. These could involve executives, public sector leaders, or regulators discussing lived experiences of decision making, oversight, and accountability, helping bridge the gap between governance theory and organisational reality. To support inclusion, hybrid and asynchronous participation formats should be embedded by design. This includes interactive digital platforms for pre-submission of ideas, live virtual breakout sessions, and post-Dialogue knowledge sharing. This reduces barriers for participants from underrepresented regions and smaller organisations. Finally, ongoing communities of practice linked to the Dialogue would sustain engagement beyond the event itself. These groups could focus on specific themes, share evolving insights, and feed evidence back into future Dialogues. Together, these formats would encourage candid exchange, practical problem solving, and sustained collaboration, strengthening the AI Dialogue's effectiveness and real-world relevance.

Below are examples of policies, practices, platforms, and approaches that demonstrate effective AI governance and offer practical solutions to current challenges. These examples are drawn from policy, organisational practice, and ecosystem-level activity rather than any single regulatory model. Policy and framework approaches • Risk-based governance frameworks that differentiate obligations based on use case, context, and impact. This approach supports safety and accountability while avoiding unnecessary constraints on low-risk innovation. • Regulatory sandboxes and testbeds that allow organisations to trial AI systems under supervision. These enable learning for both regulators and users, and reduce uncertainty around compliance and risk management. • AI assurance and auditing frameworks that translate high-level principles into assessable controls, documentation, and assurance processes. Organisational practices • Internal AI governance committees or boards that bring together technical, legal, ethical, and business perspectives. These structures help embed accountability and human oversight into decision-making. • Model and data documentation practices, such as model cards and data impact assessments, which improve transparency, traceability, and organisational learning. • Responsible procurement policies that require suppliers to meet defined standards on safety, explainability, and risk management. Platforms and ecosystem approaches • University and business school-led programmes that combine executive education, applied research, and practitioner engagement to build AI governance capability at leadership and workforce levels. • Regional AI hubs and innovation intermediaries that support SMEs through guidance, shared tools, and peer learning, helping translate governance expectations into practical action. • Multi-stakeholder partnerships and communities of practice that enable ongoing knowledge exchange and communication across sectors and jurisdictions. Together, these approaches demonstrate that effective AI governance is not achieved through regulation alone, but through coordinated efforts across policy, education, organisational practice, and regional ecosystems. The most successful solutions are those that are practical, proportionate, and capable of evolving alongside technological change.