CBS AI

A successful first Global Dialogue on AI Governance should create practical alignment between policy, education, industry, civil society, and youth around how AI can be adopted responsibly and inclusively. From our perspective, success would mean three outcomes. First, a stronger shared understanding of what trustworthy AI requires in practice: not only principles, but concrete approaches to reliability, transparency, human oversight, accountability, and risk management. Second, a clear commitment to AI capacity-building, especially for students and young professionals. Many people can now use AI tools, but far fewer are trained to evaluate outputs, understand limitations, design responsible workflows, and translate AI into real-world value. Closing this capability gap is essential if AI governance is to become operational rather than theoretical. Third, the Dialogue should help bridge the gap between global governance discussions and local implementation. AI affects labour markets, education, business transformation, access to opportunity, culture, language, and trust in institutions. A successful dialogue should therefore make space for diverse voices, including students, educators, entrepreneurs, and communities who are directly affected by AI but often underrepresented in governance conversations. Ultimately, success would be a dialogue that turns shared concern into shared capability: clearer standards, stronger cooperation, and more people equipped to build and use AI responsibly.

• AI capacity-building
• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI

These priorities reflect our focus on responsible AI capability-building. AI governance cannot remain only at the level of high-level principles. It must translate into practical skills, institutional readiness, and responsible implementation. AI capacity-building is central because students and future professionals need to understand not only how to use AI, but how to evaluate reliability, identify risks, design workflows, and apply human judgment. Safe, secure and trustworthy AI is equally important because adoption without trust, security, and reliability can create harm and reduce public confidence. We also selected transparency, accountability, and human oversight because AI systems should remain understandable, contestable, and governed by clear responsibility structures. Finally, the social, economic, ethical, cultural, linguistic, and technical implications of AI matter because AI governance must reflect real-world effects on people, work, education, opportunity, inclusion, and democratic participation.

One important cross-cutting issue is the gap between AI policy and implementation capability. Many AI governance discussions focus on principles, regulation, and risk categories, but less attention is given to how people actually develop the skills to implement AI responsibly in real organizations, classrooms, and communities. This includes practical competencies such as evaluating AI outputs, designing human review processes, measuring value, documenting decisions, and knowing when not to automate. Another emerging issue is the role of young people and students as active governance stakeholders. Students are often treated as future users or future workers, but they are already using AI to learn, create, build, and make decisions. Their perspectives should be included more directly in governance conversations, especially because today's students will soon shape AI adoption in business, public institutions, and civil society. A final issue is trust in AI-mediated environments. As AI-generated content, synthetic communication, and automated decisions become more common, societies will need better norms for authenticity, disclosure, accountability, and human connection. Governance should therefore address not only technical safety, but also the human conditions needed for trust

In Europe and Denmark, the most significant governance gap is no longer awareness of AI risks, but the ability to translate governance into practical implementation. The EU AI Act and related initiatives create an important framework for trustworthy AI, but many organizations, educators, and young professionals still lack the operational capabilities needed to apply these rules in practice. This affects our sector directly. Students and early-career professionals are already using AI in learning, work, research, communication, and entrepreneurship, but often without enough training in reliability assessment, data protection, transparency, human oversight, or responsible workflow design. This creates several challenges. First, there is a skills gap between AI use and AI governance readiness. Second, smaller organizations and student-led initiatives may struggle to understand what responsible compliance and good practice look like in concrete terms. Third, uneven access to AI literacy may widen existing inequalities between those who can use AI strategically and those who only use it superficially. At the same time, there is a major opportunity. Denmark and Europe can become leaders in responsible AI adoption by investing in capacity-building, applied education, and cross-sector collaboration between universities, companies, policymakers, and civil society. If governance is connected to practical training, students can become not only AI users, but responsible implementers who understand business value, risk, accountability, and human impact. The opportunity is to make AI governance teachable, usable, and actionable. This would help turn regulation into real-world trust, innovation, and inclusive capability-building.

The AI Dialogue can play a crucial role as a bridge between fragmented AI governance efforts. At present, many countries, regions, companies, universities, and civil society actors are developing AI rules, principles, and practices in parallel. This creates momentum, but also fragmentation. The Dialogue can help by creating a trusted international space where different approaches are compared, lessons are shared, and common expectations are developed without assuming that one model fits all contexts. Its most important role should be to move AI governance from abstract principles toward practical cooperation. This means supporting shared language around trustworthy AI, human rights, transparency, accountability, safety, and capacity-building. It also means helping countries and sectors learn from each other about implementation: what works, what fails, and what support is needed. The Dialogue should also ensure that voices beyond governments and large technology companies are included. Students, educators, researchers, SMEs, workers, civil society organizations, and affected communities should have meaningful channels to contribute. AI governance will only be legitimate if it reflects the people and institutions directly shaped by AI adoption. Finally, the Dialogue can help reduce global inequality in AI development and use. By prioritizing capacity-building, open knowledge exchange, and inclusive participation, it can support countries and communities that otherwise risk being left behind. In short, the AI Dialogue should become a practical cooperation platform: connecting governance, evidence, education, and implementation.

The AI Dialogue should build on existing initiatives rather than duplicate them. Important foundations include the Global Digital Compact, the Independent International Scientific Panel on AI, UNESCO's Recommendation on the Ethics of AI, the OECD AI Principles and GPAI partnership, the Council of Europe Framework Convention on AI, the EU AI Act, and regional, national, academic, and civil society initiatives on responsible AI. These already provide important work on human rights, safety, transparency, accountability, ethics, technical standards, and policy learning. UNESCO's recommendation is a global ethics standard, while the Council of Europe Convention is the first legally binding international treaty in this field. The added value of the AI Dialogue should be coherence, inclusion, and translation into practice. First, it can connect initiatives that currently operate in separate policy, technical, academic, and industry spaces. Second, it can create a broader participation channel for actors who are often underrepresented, including students, young professionals, educators, SMEs, and communities affected by AI systems. Third, it can help translate principles into operational capability by focusing on implementation, capacity-building, and shared learning. For our sector, the most valuable contribution would be connecting governance with education. Universities and student-led initiatives can help turn AI governance into practical skills: evaluating outputs, designing human oversight, understanding risks, documenting decisions, and building responsibly. The Dialogue should therefore act as a coordination layer: not replacing existing mechanisms, but helping them become more interoperable, accessible, and useful for real-world AI adoption

Different stakeholders can contribute by bringing evidence from their own context and helping turn AI governance into practical implementation. Governments can share regulatory approaches, public sector needs, and lessons from implementation. Industry can contribute technical expertise, risk management practices, and real-world examples of AI deployment. Academia can provide independent research, evaluation methods, and education models. Civil society can represent affected communities, human rights concerns, and social impact. Students, young professionals, workers, educators, SMEs, and local communities should also be included because they experience AI adoption directly in learning, work, and daily life. The AI Dialogue should be structured as more than high-level plenary discussions. It should include thematic working groups, regional consultations, youth and student forums, expert panels, public submissions, and practical implementation labs. These formats would allow different stakeholders to move from broad principles to concrete questions such as: how to assess risk, how to design human oversight, how to build AI literacy, how to protect rights, and how to ensure inclusion. To be meaningful, participation should be accessible, multilingual, and transparent. Contributions should be documented publicly where possible, and participants should be able to see how their input shapes outcomes. The strongest format would combine global coordination with local relevance: international dialogue, regional listening sessions, and practical case-based working groups that produce usable guidance for policymakers, educators, companies, and communities.

Several voices remain underrepresented in global AI governance discussions. These include young people and students, workers affected by automation, educators, SMEs, local public sector practitioners, civil society organizations, disability communities, Indigenous communities, linguistic minorities, low-resource language communities, and people from countries or regions with limited access to AI infrastructure. Many discussions are still shaped mainly by governments, large technology companies, and technical experts, even though AI affects many more people. Students and young professionals are especially important because they are already using AI in education, work, research, entrepreneurship, and communication. They should not only be treated as future users or future workers, but as current stakeholders with direct experience of how AI changes learning and opportunity. These groups could be included through dedicated youth consultations, community roundtables, regional listening sessions, open written submissions, multilingual participation channels, and funded access for participants who would otherwise not be able to join. The Dialogue should also create formats that are not overly technical, so that lived experience and practical insight can contribute alongside expert knowledge. Inclusion should not mean symbolic representation only. Underrepresented groups should have clear channels to shape priorities, comment on draft outputs, and participate in working groups. This would make AI governance more legitimate, grounded, and responsive to real social impact.

The AI Dialogue should use engagement formats that move beyond speeches and allow participants to work with real governance dilemmas. One effective format would be case-based implementation labs, where mixed stakeholder groups examine concrete AI scenarios and identify risks, accountability gaps, human oversight needs, and practical safeguards. This would make governance more actionable. Another useful format would be youth and student assemblies, where young people can share how AI is already changing education, career preparation, entrepreneurship, and everyday decision-making. These assemblies should feed directly into formal dialogue outputs. The Dialogue could also use multi-stakeholder roundtables with deliberately mixed groups, including government, industry, academia, civil society, students, workers, and affected communities. This would reduce siloed discussion and surface trade-offs more clearly. A fourth format could be policy-to-practice clinics, where organizations bring real implementation questions and receive structured feedback from legal, technical, ethical, and social impact perspectives. Finally, digital participation should be designed carefully. Online submissions, multilingual surveys, interactive mapping of AI governance challenges, and public comment periods can widen access, especially for those unable to attend physically. The most important principle is that engagement should produce visible outputs: draft recommendations, implementation checklists, case libraries, capacity-building needs, and stakeholder commitments. Dynamic engagement happens when people can see that their contribution shapes the final direction.

Effective AI governance should combine regulation, technical standards, organizational practice, and education. One important example is the EU AI Act, which uses a risk-based approach and creates obligations around prohibited practices, high-risk AI systems, transparency, and general-purpose AI. This is valuable because it links governance duties to the level of potential harm. Another useful approach is the NIST AI Risk Management Framework, which helps organizations identify, measure, manage, and govern AI risks across the AI lifecycle. Its value is that it can be translated into practical internal processes, not only legal compliance. UNESCO's Recommendation on the Ethics of AI and the OECD AI Principles also provide important foundations by emphasizing human rights, transparency, accountability, inclusion, safety, and democratic values. At the practical level, strong governance practices include AI impact assessments, model documentation, clear human oversight responsibilities, audit trails, incident reporting, procurement standards, and regular testing for bias, robustness, privacy, and security risks. Regulatory sandboxes are also valuable because they allow companies, public institutions, researchers, and regulators to test AI systems in controlled environments before wider deployment. Finally, applied capacity-building should be treated as a governance tool. Students, workers, and professionals need training in evaluating AI outputs, identifying risks, designing responsible workflows, and knowing when not to automate. Good governance is not only about rules. It is about building the capability to apply them responsibly in real contexts.