FARI - AI for the Common Good

In my view, the first Global Dialogue on AI Governance would be a success if it achieves three things at once: legitimacy, usefulness, and continuity. First, it should demonstrate that this is a genuinely inclusive multilateral space where all Member States, alongside the private sector, civil society, academia, the technical community and international organizations, can contribute meaningfully to the conversation on AI governance. That inclusiveness and interoperability is a core part of the Dialogue's UN mandate and one of its main sources of added value. Second, it should produce practical outcomes, not only general statements. A strong first outcome would be a short action-oriented summary identifying a small number of priority cooperation areas, such as capacity-building, interoperability across governance approaches, public-interest use cases, and mechanisms to share lessons from implementation. This would respond to a major gap in today's landscape: many countries remain underrepresented in international AI governance efforts, while implementation capacity is uneven. Third, it should establish a clear bridge between science and policy. The first Dialogue should show that the Scientific Panel's report is not simply received, but actively used to inform policy discussion, identify shared risks and opportunities, and guide cooperation priorities. That science-policy connection is one of the reasons the Dialogue and the Panel were created together. Ultimately, success would mean that participants leave Geneva not only with a better shared understanding of AI governance challenges, but with a credible pathway for continued cooperation before the next Dialogue.

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

Our selection reflects FARI's role as a Brussels-based academic institute on AI for the common good, working at the intersection of research, public-sector innovation, and multi-stakeholder engagement. FARI's mission explicitly focuses on the adoption and governance of AI, data and robotics in an inclusive, ethical and sustainable way, with emphasis on explainable and trustworthy AI, open data, and human-centric approaches. We therefore prioritize safe, secure and trustworthy AI because trust, safety and responsible deployment are foundational for meaningful public adoption. We prioritize AI capacity-building because governance cannot be effective if public administrations, local ecosystems and smaller actors lack the skills, institutional readiness and implementation support needed to operationalize it. We prioritize the social, economic, ethical, cultural, linguistic and technical implications of AI because Brussels is a particularly relevant environment: institutionally complex, multilingual and internationally diverse, making it essential to understand how AI governance must respond to varied local realities. Finally, we prioritize interoperability of governance approaches because this is one of the most pressing practical challenges in today's landscape. Even when actors share common goals, local, regional and national contexts often shape implementation differently. For us, the value of the Global Dialogue lies in helping bridge these differences, so that AI governance can be discussed globally while remaining implementable locally and aligned internationally.

1) Implementation and institutional readiness should be treated as a distinct governance issue. In practice, many challenges emerge not at the level of principles, but when public institutions try to translate them into procurement rules, risk assessment, oversight processes, skills development, and day-to-day operational use. For a Brussels-based institute working with public actors, this "last mile" of governance is often where fragmentation becomes most visible. FARI's own mission is rooted in fostering the adoption and governance of AI, data and robotics in an inclusive, ethical and sustainable way, including at local level. 2) Data governance across the AI lifecycle merits stronger visibility. This goes beyond open data alone and includes data quality, provenance, access conditions, interoperability, stewardship, and the governance of shared or sensitive data environments. In many real-world settings, AI governance is only as strong as the data governance behind it. 3) Concentration of infrastructure and dependency risks should be recognised more explicitly. Capacity-building is essential, but it is also important to address unequal access to compute, cloud infrastructure, foundation models, and technical talent, since these structural asymmetries shape who can meaningfully participate in AI development and governance.

In the Brussels-Capital Region and in our public-interest research and public-sector innovation context, the main effect of recent AI governance developments is that AI is moving from exploration to implementation. The EU AI Act has created a clearer common baseline for risk, safety and responsibilities, while the regional digital ecosystem is also becoming more active around AI experimentation, guidance and data strategy. In Brussels, Paradigm coordinates the Region's digital strategy and data governance, has run exploratory public-service AI pilots, and has issued internal generative AI guidance for administrations. This creates a major opportunity: public institutions now have a stronger foundation to test AI in a more structured way, build internal capacity, and align innovation with public values. For FARI, our mission being to foster AI, data and robotics in a trustworthy, open, inclusive and responsible way, this opens space for applied collaboration between academia, administrations, industry and citizens. The most significant challenge is that implementation capacity remains uneven. Governance gaps appear when administrations must translate high-level principles into procurement, risk assessment, oversight, skills development and day-to-day operational practice. In addition, interoperability remains difficult: the current European data and regulatory landscape is still perceived as fragmented and complex to interpret, which can slow adoption and create siloed approaches across institutions and projects due to their intrinsic diversities. For Brussels specifically, these issues are amplified by an institutionally complex and multilingual environment. This makes the region a valuable testbed, but also means that trust, inclusion, linguistic accessibility and coordination across stakeholders must be addressed more deliberately. Overall, the opportunity is significant, but success depends on turning governance from compliance language into practical institutional readiness.

The AI Dialogue can play an important role by creating a genuinely inclusive space where international cooperation on AI governance becomes more practical and less fragmented. At the moment, many countries, regions and institutions are moving forward through different legal, technical and policy approaches, which makes alignment difficult even when the overall goals are similar. The value of the Dialogue is that it can help connect these approaches, not by forcing uniformity, but by improving mutual understanding, interoperability and shared learning. It can also help strengthen the link between scientific evidence and policy discussion, so that governance debates are informed by a clearer understanding of both risks and opportunities. For actors working in implementation contexts, such as public administrations, cities, and research institutes like FARI, the Dialogue is especially valuable if it supports practical cooperation: peer exchange, capacity-building, and examples of what responsible implementation looks like in different settings. Its success will depend on whether it moves beyond general discussion and helps build sustained cooperation that is globally informed but locally usable.

From our perspective, the AI Dialogue should build on existing frameworks and communities rather than duplicate them. In particular, it should connect with the UNESCO Recommendation on the Ethics of AI, which remains the first global normative standard in this field; the OECD AI Principles and the Global Partnership on AI (GPAI), which already provide a strong basis for multi-stakeholder exchange on trustworthy AI; and the Council of Europe Framework Convention on AI, which is the first legally binding international treaty on AI. It should also stay connected to implementation-oriented regional efforts, such as the EU AI Act, and to broader UN platforms like AI for Good, which already convene actors around standards, skills and applied use cases. The added value of the AI Dialogue is therefore not to create another standalone governance track, but to act as the bridge between these initiatives. Its strength lies in the UN's universal convening role: it can bring together countries and stakeholders that are not equally represented in existing forums, while helping translate fragmented principles and regional experiences into a more inclusive international conversation.