University of Wollongong in Dubai

A successful first Global Dialogue on AI Governance should move beyond high-level consensus toward practical, implementable outcomes that can guide diverse stakeholders across regions and sectors. First, the Dialogue should advance a shared understanding of trustworthy AI, supported by a lifecycle-based perspective that spans data, model development, deployment, monitoring, and governance. This would help align fragmented approaches and ensure accountability is embedded throughout the system. Second, it should establish a global repository of tested practices, policy tools, and case studies, including contributions from underrepresented regions. This would enable knowledge sharing, support context-sensitive adaptation, and reduce duplication of effort. Third, success should be reflected in capacity-building that goes beyond infrastructure, incorporating ethical literacy, critical thinking, and responsible use of AI across domains such as education, healthcare, and public services. Fourth, the Dialogue should produce actionable and measurable recommendations within the Co-Chairs' Summary, allowing progress to be tracked and evaluated between iterations. Finally, a clear mechanism for continuity and sustained engagement is essential. This could include thematic working groups, structured stakeholder input processes, and ongoing platforms for collaboration, ensuring that insights from the Dialogue are systematically captured and translated into practice.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

These priorities reflect the need to move from high-level principles toward operational, context-sensitive AI governance. "Safe, secure and trustworthy AI" is foundational, but trust must be clearly defined and implemented across the full AI lifecycle, from data to deployment and ongoing monitoring. Without this, governance risks remaining fragmented and difficult to apply in practice. "Transparency, accountability, and human oversight" are essential to operationalising trust. They ensure that AI systems are explainable, auditable, and aligned with human values, particularly in high-stakes domains such as healthcare and education. The inclusion of "social, economic, ethical, cultural, linguistic and technical implications" recognises that AI governance cannot be effectively standardised without considering local contexts. Cultural norms, language diversity, and socio-economic realities significantly shape how AI systems are experienced and trusted. "Protection and promotion of human rights" ensures that governance frameworks remain grounded in fundamental values, particularly as AI systems increasingly influence access to services, opportunities, and decision-making processes. Together, these priorities support a holistic approach to AI governance, balancing technical robustness with social responsibility and ensuring that trust, accountability, and inclusivity are embedded across all stages of AI development and use.

One key cross-cutting issue is the operationalisation and measurement of trust in AI systems. While trust is implicit across several themes, there is limited emphasis on how trustworthiness can be consistently defined, assessed, and monitored across different contexts and stages of the AI lifecycle. Developing shared indicators and practical frameworks for evaluating trust would significantly strengthen governance efforts. A second emerging issue is the need to integrate education, behaviour, and human interaction with AI systems into governance discussions. Current approaches often focus on regulation and technical standards, but long-term effectiveness depends on how individuals understand, engage with, and use AI responsibly. Embedding ethics within education systems, professional training, and everyday use contexts is essential for sustainable governance. Additionally, there is a need for more emphasis on structured stakeholder engagement methodologies. While inclusivity is prioritised, the lack of systematic approaches to capturing and analysing stakeholder input can limit the depth, comparability, and usability of insights generated through consultations. Finally, global power asymmetries in AI development and governance remain a critical issue. There is a risk that dominant regions shape standards that may not align with the needs, values, and realities of other parts of the world. Ensuring equitable representation and enabling context-sensitive adaptation should be a central consideration across all themes.

Across parts of the Global South rapid advancements in AI development are creating both significant opportunities and emerging governance gaps. One key challenge is the mismatch between the pace of AI adoption and the maturity of governance frameworks, particularly at the operational level. While high-level policies and national strategies are progressing, there remains limited clarity on how principles such as trust, accountability, and transparency are implemented consistently across sectors such as education and healthcare. A second challenge relates to capacity-building gaps, not only in technical expertise but in ethical literacy and responsible use. As AI tools become widely accessible, users across sectors often engage with these systems without sufficient understanding of risks, limitations, or implications, which can undermine trust and lead to unintended consequences. At the same time, there are important opportunities. The UAE and similar regions are well-positioned to leapfrog legacy systems and embed governance considerations early in AI adoption. There is growing openness to innovation, cross-sector collaboration, and experimentation with emerging technologies. Another opportunity lies in developing context-sensitive governance approaches that reflect linguistic, cultural, and socio-economic realities, rather than adopting models developed in different regions without adaptation. Finally, increased investment in education, research, and multi-stakeholder engagement presents an opportunity to build governance frameworks that are not only technically robust but also socially grounded. Addressing these gaps while leveraging these opportunities will be critical to ensuring that AI development is both trustworthy and inclusive across diverse contexts.

The Global Dialogue on AI Governance can play a critical role in advancing international cooperation by moving beyond general alignment toward practical coordination and shared implementation approaches. First, it can serve as a platform to develop a common language around trustworthy AI, enabling stakeholders across regions to align on key concepts such as transparency, accountability, and human oversight, while still allowing for context-sensitive adaptation. Second, the Dialogue can facilitate knowledge exchange through structured sharing of tested practices, policy tools, and case studies, particularly from underrepresented regions. This would help bridge gaps between countries at different stages of AI development and reduce duplication of effort. Third, it can support capacity-building through collaborative mechanisms, including partnerships between countries, institutions, and sectors. This is especially important for ensuring that developing regions are not only adopters of AI technologies but active contributors to governance frameworks. Fourth, the Dialogue can promote interoperability of governance approaches by identifying areas of convergence across existing frameworks, while acknowledging differences in regulatory, cultural, and socio-economic contexts. Finally, its most important role may be in establishing continuity and accountability in global cooperation. By creating ongoing thematic working groups, structured stakeholder engagement processes, and mechanisms to track progress over time, the Dialogue can ensure that international cooperation evolves from periodic discussion into sustained, collective action.

Key initiatives include those led by UNESCO on AI ethics, the OECD AI Principles, and platforms such as the AI for Good Global Summit, which already facilitate dialogue between governments, industry, and academia. At the regional level, forward-looking frameworks such as the Dubai Universal Blueprint for Artificial Intelligence, initiatives by Digital Dubai, and emerging regulatory approaches such as the UAE AI Act demonstrate how governance can be embedded within national innovation strategies. The added value of the Global Dialogue lies not in creating new principles, but in connecting, aligning, and operationalising existing ones. It can serve as a coordinating platform that identifies areas of convergence across frameworks, supports interoperability, and translates high-level principles into practical guidance. Importantly, the Dialogue can strengthen inclusion by amplifying perspectives from underrepresented regions and sectors, ensuring that governance frameworks reflect diverse realities rather than a limited set of contexts. It can also add value by establishing mechanisms for structured knowledge exchange, such as a shared repository of case studies, implementation tools, and lessons learned, enabling stakeholders to move from discussion to practice. Finally, by linking existing initiatives through ongoing working groups and collaborative networks, the Dialogue can help shift global efforts from fragmented activities toward sustained, coordinated action in AI governance.

Effective stakeholder contribution requires moving beyond open dialogue toward structured, inclusive, and comparable engagement. First, the Dialogue should incorporate guided participation formats, such as scenario-based discussions, role-based simulations, and structured trade-off exercises. These approaches enable stakeholders from diverse backgrounds to engage meaningfully, even without deep technical expertise, while generating more actionable and comparable insights. Second, participation should be multi-layered, combining: -High-level plenaries for strategic direction -Thematic breakout sessions for focused discussion -Smaller working groups tasked with producing specific outputs Third, the Dialogue should enable continuous engagement beyond the event through digital platforms that allow stakeholders to contribute asynchronously, particularly those unable to participate in real time due to geographic or resource constraints. Fourth, mechanisms should be introduced to systematically capture and synthesise stakeholder input, ensuring that contributions are not only heard but translated into policy-relevant outputs. Finally, clear roles should be defined for different stakeholders, including governments, industry, academia, and civil society, ensuring that each group contributes according to its expertise while maintaining balanced representation. This approach would strengthen both the quality and inclusivity of participation, while ensuring that the Dialogue produces actionable outcomes.

Several important voices remain underrepresented in global discussions on AI governance. First, there is limited representation from the Global South, particularly practitioners and communities directly affected by AI deployment. While national perspectives may be present, local and sector-specific experiences are often underrepresented. Second, educators, students, and everyday users of AI systems are rarely included in governance discussions, despite being among the most impacted groups. Their perspectives are critical for understanding how AI is actually used, misused, and trusted in real-world contexts. Third, non-technical stakeholders, including professionals in healthcare, education, and public services, are often excluded due to highly technical formats of engagement. Fourth, linguistically and culturally diverse communities remain underrepresented, particularly where language barriers limit participation in global forums. To address this, the Dialogue should: -Adopt inclusive participation formats that do not rely solely on technical expertise -Provide multilingual engagement and documentation -Enable remote and asynchronous participation -Partner with regional networks to bring in local stakeholders -Create pathways for youth and early-career contributors to engage meaningfully Ensuring broader representation will not only improve inclusivity but also lead to more grounded, context-sensitive, and trustworthy governance outcomes.

o foster meaningful and dynamic engagement, the Dialogue should move beyond traditional panel formats and adopt participatory, scenario-driven approaches that allow stakeholders to actively engage with real-world governance challenges. One effective format is the use of governance sandboxes, where participants work through simulated AI deployment scenarios across sectors such as healthcare or education. In these controlled environments, stakeholders can explore trade-offs, test governance responses, and better understand the implications of decisions across the AI lifecycle. Complementing this, design thinking workshops can be integrated into the Dialogue to structure engagement around key stages such as problem definition, ideation, and solution development. This approach enables diverse participants, including non-technical stakeholders, to contribute meaningfully while ensuring discussions remain focused and outcome-oriented. Another valuable format is role-based simulation exercises, where participants represent different stakeholders (e.g., regulators, developers, users, policymakers) and engage in guided decision-making processes. This helps surface competing priorities, ethical tensions, and practical constraints in a structured way. To enhance inclusivity and continuity, these formats can be supported through hybrid and asynchronous participation mechanisms, allowing stakeholders to contribute beyond the live sessions and across time zones. Finally, outputs from these sessions should be systematically captured and synthesised into actionable insights and policy-relevant recommendations, ensuring that engagement directly informs governance development. Such formats would transform the Dialogue from a platform for discussion into a space for experimentation, learning, and practical co-creation in AI governance.

Effective AI governance is strengthened when high-level principles are translated into practical, testable, and context-sensitive approaches. One promising approach is the use of governance sandboxes, where AI systems are evaluated in controlled, scenario-based environments before large-scale deployment. These sandboxes enable stakeholders to assess risks, explore trade-offs, and test governance responses across the AI lifecycle in a structured and collaborative manner. Complementing this, design thinking-based policy development has shown strong potential in bridging the gap between principles and practice. For example, in recent multi-stakeholder engagements under the Yathiqu project (funded by Dubai Future Foundation), design thinking methodologies were used to bring together participants from healthcare, policy, and technical domains. Through structured stages such as problem framing, stakeholder mapping, and iterative solution development, participants engaged with realistic AI deployment scenarios, enabling them to surface ethical tensions, identify governance gaps, and co-create context-sensitive responses. At the international level, frameworks developed by UNESCO and the OECD provide important ethical foundations. However, their effectiveness is enhanced when paired with implementation-focused tools, such as lifecycle-based governance models, audit mechanisms, and sector-specific guidelines. In practice, embedding governance within education and professional training further strengthens impact by promoting ethical awareness and responsible AI use among developers and users. Together, these approaches demonstrate that effective AI governance requires not only principles, but also mechanisms for experimentation, stakeholder engagement, and continuous learning.