Brunel University of London

A successful first Global Dialogue on AI Governance should be measured by its ability to move beyond high-level convergence and generate practical mechanisms that strengthen implementation capacity and interoperability across governance systems. A primary outcome would be clearer articulation of how different regulatory and institutional approaches can interact in practice, particularly across risk-based, principles-based, and sector-specific regimes. Given increasing regulatory divergence globally, success would depend on identifying translation points that allow systems to remain context-sensitive while enabling cross-border coherence. A further important outcome would be progress in addressing the implementation gap in AI governance, particularly the mismatch between rapidly evolving AI systems and the slower development of institutional, technical, and workforce capacity required for effective oversight. This includes strengthening regulatory readiness, evaluation capability, and organisational competence in deploying and monitoring AI systems across sectors. The Dialogue should also help clarify expectations around operational governance functions, particularly post-deployment monitoring, auditability, and accountability mechanisms. These areas remain underdeveloped in many jurisdictions despite being central to real-world risk management. In addition, success would be reflected in stronger recognition of the need for governance interoperability frameworks, enabling alignment not through uniform regulation, but through shared reference structures, methodological compatibility, and mutual recognition approaches. Success would also be evident if the Dialogue contributes to reframing AI governance as an implementation and capacity challenge rather than solely a normative or principles-based exercise, ensuring that governance frameworks are better aligned with institutional realities and sectoral constraints across different national contexts.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• AI capacity-building
• Interoperability of governance approaches

These three thematic areas were selected as they reflect my research and policy-oriented work on AI as a socio-technical and economic system, where outcomes are shaped not only by technological design but also by institutional capacity, market structure, and regulatory coherence across jurisdictions. AI capacity-building is central to this perspective, particularly in relation to productivity, innovation diffusion, and labour market transformation. Small and medium-sized enterprises (SMEs) play a critical role in economic growth and employment generation globally; however, their ability to adopt and effectively integrate AI is often constrained by limited technical capability, organisational readiness, and access to skilled labour. Strengthening capacity in this domain is therefore essential to ensuring inclusive participation in AI-driven economic transformation and mitigating widening disparities in productivity and competitiveness. The second focus captures the social, economic, ethical, cultural, linguistic, and technical implications of AI, which are increasingly evident across sectors undergoing rapid digital transformation. From an economic governance perspective, particular attention must be given to issues such as overreliance on automated systems, the reconfiguration of decision-making processes, and the distributional consequences of AI adoption across labour markets and public services. The third area, interoperability of governance approaches, reflects the growing fragmentation of regulatory regimes across regions. Divergent governance models risk creating inefficiencies, regulatory arbitrage, and unequal access to technological benefits. This is closely linked to broader structural concerns regarding data ownership, infrastructure concentration, and the decentralisation of digital power, where a limited number of firms and jurisdictions increasingly shape global innovation trajectories. Such concentration raises important questions regarding economic resilience, market competition, and long-term governance stability. Collectively, these priorities reflect the need for economically grounded, institutionally aware, and internationally interoperable governance frameworks, and align closely with my interest in contributing to high-level policy advisory and consultancy work with international organisations, including the United Nations.

While the listed themes provide a comprehensive foundation for AI governance, several cross-cutting and emerging issues warrant further attention due to their systemic economic and institutional implications. A key gap relates to the political economy of AI infrastructure and data governance, particularly the increasing concentration of data, computational resources, and model development within a limited number of firms and jurisdictions. This structural asymmetry shapes global innovation trajectories, affects market competition, and raises concerns regarding dependency and the unequal distribution of value generated by AI systems. These dynamics are not fully captured within existing thematic areas but significantly influence their practical implementation. A second emerging issue concerns systemic labour market adjustment and productivity transformation, particularly in relation to SMEs and non-technology sectors. While capacity-building is recognised, there is limited explicit attention to coordinated reskilling, redeployment strategies, and institutional readiness for widespread AI diffusion. This is critical for ensuring inclusive productivity gains and avoiding uneven adjustment costs across economies. A further issue is the increasing institutional reliance on AI in decision-making processes, which extends beyond transparency and oversight to questions of long-term substitution of human judgement in economic, public, and regulatory systems. This raises broader concerns about accountability diffusion and systemic risk propagation. Finally, governance interoperability in practice remains a cross-cutting challenge. Divergent regulatory models across jurisdictions risk creating compliance fragmentation and unequal access to AI capabilities, particularly affecting SMEs and emerging economies. Addressing these issues requires greater integration between economic governance, institutional capacity-building, and global coordination mechanisms to ensure that AI development supports inclusive, resilient, and balanced innovation systems.

In the UK context and across comparable advanced and emerging economies, governance gaps in AI are increasingly shaping both sectoral outcomes and broader patterns of economic adjustment, with particularly visible effects in knowledge-intensive and creative industries. A key challenge arises from uneven implementation of governance frameworks relative to the speed of AI adoption. While the UK and EU are developing regulatory approaches—ranging from principles-based governance to the EU AI Act's risk-based model—organisational readiness often lags behind. Many firms, particularly SMEs, lack the technical, legal, and institutional capacity to operationalise governance requirements in practice, limiting their ability to adopt AI safely and effectively. In parallel, there is growing evidence of labour market disruption and organisational restructuring, especially in sectors such as gaming, media, and digital content production. AI adoption has contributed to workforce displacement in some contexts, while in others it has led to inefficient implementation cycles, including partial reversals where firms have reintroduced human roles following productivity or quality losses. This reflects both the opportunities and risks of rapid, uneven diffusion. A further challenge relates to data concentration and infrastructure dependency, where a small number of global technology providers shape access to models, datasets, and computational resources. This raises concerns regarding strategic dependence, bargaining asymmetries, and reduced policy autonomy for national regulators and firms. At the same time, there are significant opportunities. The UK's research and innovation ecosystem, combined with emerging regulatory frameworks, positions it to develop high-quality governance standards and applied AI assurance practices, particularly in high-impact domains such as health, finance, and public services. There is also scope for SMEs to become key beneficiaries of well-designed capacity-building initiatives, enabling wider productivity gains and innovation diffusion. These developments collectively indicate that the most significant impact of current governance gaps is uneven institutional capacity to absorb, regulate, and benefit from AI at scale.

The AI Dialogue can play a critical role in advancing international cooperation by addressing structural gaps in current AI governance, particularly those related to fragmentation, capacity asymmetries, and the concentration of technological resources. A central contribution would be to support practical interoperability between governance approaches. As regulatory models diverge, there is a need for mechanisms that enable alignment in implementation rather than convergence at the level of principles alone. This is particularly relevant in sectors such as healthcare, where AI-enabled diagnostic systems may be developed in one jurisdiction and deployed in another, requiring consistent approaches to validation, safety, and accountability. Similarly, in financial services, cross-border AI applications in credit scoring and fraud detection require interoperable standards to avoid regulatory inconsistency. The Dialogue can also play an important role in addressing disparities in institutional and economic capacity. SMEs across sectors such as manufacturing, retail, and professional services often lack the resources to implement governance requirements effectively, limiting their participation in AI-driven productivity gains. International cooperation should therefore prioritise capacity-building that includes regulatory implementation, workforce development, and organisational readiness. In addition, the Dialogue provides an opportunity to engage more directly with the political economy of AI. In areas such as cloud computing and large-scale model development, dependence on a small number of global providers raises concerns about access, pricing, and strategic autonomy for both firms and governments. The Dialogue can also help align AI governance with labour market transitions, for example in creative industries and digital content sectors, where AI adoption is reshaping production processes and employment structures. Through these functions, the AI Dialogue can support a more coherent and implementation-focused model of international cooperation.

Existing initiatives provide important building blocks for AI governance, particularly in regulatory development, capacity-building, and economic strategy, but they remain fragmented in application and uneven in impact. Regulatory approaches such as those emerging in the European Union and the United Kingdom illustrate different models of governing AI. While these frameworks provide useful guidance, their divergence creates practical challenges in sectors such as healthcare, where clinical AI tools must meet varying regulatory expectations, and in financial services, where cross-border AI systems must navigate inconsistent compliance requirements. This fragmentation can be particularly burdensome for SMEs operating across multiple jurisdictions. In parallel, investments in AI infrastructure and innovation ecosystems- such as national compute strategies and regional AI development programmes- highlight the importance of linking governance with economic capability. However, access to these resources remains uneven, affecting sectors such as advanced manufacturing, logistics, and public services, where adoption depends on both infrastructure and institutional capacity. There are also lessons from adjacent domains, particularly in data governance and regulated data-sharing systems. For example, structured data-sharing frameworks in finance have demonstrated how standardisation, interoperability, and accountability mechanisms can support innovation while maintaining oversight. Similar approaches may be relevant for AI training data and cross-border data access. The added value of the AI Dialogue lies in addressing the gaps between these initiatives. By supporting alignment at the level of implementation, facilitating shared governance practices, and improving accessibility for underrepresented actors, the Dialogue can help ensure that existing efforts contribute to a more coherent and inclusive global AI governance system.

Effective stakeholder contribution to the AI Dialogue depends on moving beyond broad consultation toward structured, role-specific engagement that reflects the diversity of expertise, capacities, and interests across sectors and regions. Different stakeholder groups should be engaged through clearly defined functions. Governments and regulators can contribute comparative insights on policy design and implementation challenges. Industry actors, including large firms and SMEs, can provide evidence on adoption constraints, compliance burdens, and innovation dynamics. Academia can support the Dialogue through independent research, evaluation methodologies, and evidence synthesis, while civil society and professional communities can highlight societal impacts, including labour, rights, and public trust considerations. To ensure meaningful participation, the Dialogue should adopt a multi-layered structure. Plenary sessions can set strategic direction, while smaller thematic working groups or task-oriented streams focus on specific issues such as capacity-building, interoperability, or sectoral implementation. These groups should be designed to produce practical outputs, including guidance, case studies, and policy-relevant recommendations. In addition, participation should be supported through hybrid and asynchronous mechanisms, enabling contributions from stakeholders who may not be able to engage in real time. Multilingual digital platforms, structured submissions, and iterative feedback processes would improve accessibility and continuity. Particular attention should be given to the inclusion of underrepresented actors, especially SMEs, practitioners, and stakeholders from emerging economies, who are often directly affected by AI deployment but less visible in governance discussions. Targeted engagement formats, including regional consultations and sector-specific roundtables, can help address this gap. Finally, the Dialogue should incorporate feedback loops and continuity mechanisms, ensuring that stakeholder contributions inform subsequent discussions and policy development. This would strengthen accountability and support the transition from dialogue to implementation.

Underrepresented voices in global AI governance include several groups whose roles are central to the functioning of AI systems, but whose perspectives are not adequately reflected in formal governance processes. Small and medium-sized enterprises (SMEs) remain underrepresented despite their importance in employment, innovation diffusion, and productivity growth. Their limited capacity to engage in regulatory processes means that governance frameworks may not fully reflect implementation constraints or adoption realities. This is particularly relevant in areas such as AI-enabled compliance and cross-border digital systems, where coordination and interoperability challenges disproportionately affect smaller actors. Initiatives such as the ATTACC project illustrate how various stakeholders including SMEs operate within complex AI-driven infrastructures that depend on coordinated governance across jurisdictions. A critically overlooked group is data creators and platform users who generate the large-scale behavioural and interactional data underpinning LLM development. These individuals and communities- through everyday use of digital platforms- produce the datasets that enable model training, yet are rarely recognised as stakeholders in AI governance discussions. This creates a structural gap between value generation and governance recognition, raising important questions about data ownership, consent, and value distribution in AI ecosystems. Communities in emerging economies and digitally underserved regions also remain underrepresented, despite being significantly affected by AI-driven systems and infrastructure asymmetries. More inclusive participation requires multilingual engagement and regionally embedded consultation mechanisms. Workers in digital labour and creative production sectors are similarly excluded from governance discussions, despite being directly affected by AI systems that reshape content generation, attribution, and remuneration structures. Addressing these gaps requires moving toward more structured and continuous participation mechanisms that integrate perspectives from the data generation and value creation layer, as well as stakeholder groups such as the labour and SME ecosystem, into global AI governance frameworks.

Effective engagement during the AI Dialogue should be designed as a structured governance and economic coordination process, rather than a traditional consultation exercise. From an economic and strategic perspective, the objective is not only participation, but the creation of mechanisms that improve implementation capacity, reduce fragmentation, and strengthen productivity outcomes across sectors and regions. A central format would be case-based policy laboratories, where stakeholders examine real-world deployment contexts to identify governance trade-offs. For example, AI-enabled clinical decision-support systems in healthcare or automated compliance tools in cross-border trade demonstrate how regulatory requirements translate into operational constraints, safety considerations, and institutional accountability challenges. Such cases allow policymakers, firms, and researchers to evaluate governance in conditions that reflect actual economic and organisational complexity. This could be complemented by multi-stakeholder implementation labs, bringing together regulators, SMEs, technology developers, and sectoral practitioners. In areas such as financial services or digital platforms, these labs would support applied problem-solving around issues such as regulatory compliance, data governance, and interoperability across jurisdictions, particularly for smaller firms with limited institutional capacity. A further strategic dimension is the role of localised engagement ecosystems. Embedding local community participation- particularly in regions affected by AI-driven infrastructure such as data centres or digital platforms- can strengthen legitimacy and ensure that AI adoption contributes to locally anchored economic development. This also reinforces social cohesion and trust in technological transformation, which are key determinants of long-term productivity and adoption success. In parallel, a shared knowledge and experience infrastructure should be developed, enabling cross-sector and cross-country exchange of implementation lessons. For example, insights from SME adoption of AI tools in manufacturing or retail can help avoid siloed development pathways and reduce duplication of regulatory and technical effort. Finally, structured foresight exercises and continuous feedback loops should be embedded into the Dialogue architecture to ensure that insights from these engagements directly inform iterative governance development. This combination of applied case-based learning, local engagement, and cross-jurisdictional knowledge exchange would support a more economically grounded, implementation-oriented, and strategically coherent AI governance ecosystem.

Effective AI governance increasingly relies on a combination of regulatory frameworks, applied infrastructures, and sectoral practices that translate principles into operational systems across different economic contexts. A key reference point is the European Union's AI Act, which introduces a risk-based approach with obligations for high-risk systems, including transparency, human oversight, and post-deployment monitoring. The United Kingdom's principles-based regulatory model complements this through a more adaptive framework, allowing governance to evolve alongside rapid technological change. In practice, these approaches are particularly relevant in sectors such as healthcare and financial services, where AI systems must meet stringent requirements for safety, accountability, and lifecycle oversight. Sectoral regulatory guidance further reinforces implementation. For example, the Medicines and Healthcare products Regulatory Agency emphasises lifecycle governance for software as a medical device, highlighting the importance of continuous validation and post-market monitoring in clinical environments. At the international level, initiatives such as the OECD AI Principles and the Global Partnership on AI provide coordination platforms for shared governance tools, capacity-building, and cross-jurisdictional alignment. Beyond regulatory frameworks, emerging applied infrastructures demonstrate how governance can be embedded directly into operational systems. For example, the ATTACC (Automated Tax Compliance for Cross-Border Trading with GRAN-IoT) project illustrates how AI and IoT-enabled systems can support automated cross-border trade compliance by reducing administrative burdens, improving transparency, and lowering risks of fraudulent or illegal trading activities. By streamlining customs and regulatory reporting processes, such systems can strengthen SME capacity, enhance efficiency in international trade, and reduce friction in cross-border economic activity. Finally, regulated data-sharing models in sectors such as financial services provide transferable lessons on interoperability, consent, and accountability. Collectively, these examples show that effective AI governance increasingly depends on integrating regulatory design with infrastructure-level implementation and economic functionality.