MatchDay Navigator (SUKA Consulting)

A successful first Global Dialogue on AI Governance would establish shared clarity on how artificial intelligence can be governed when deployed in public‑interest and civic systems, where accountability, operational reliability, and public trust are essential. In such contexts, success would include: Clear recognition that not all AI systems are designed for consumer use, and that AI supporting mobility, public safety, event coordination, and civic services must operate with narrow scope, clearly defined purpose, and explicit sunset conditions. Inclusion of practical deployment experience from environments where systems operate under real‑world pressure and governance constraints, ensuring policy discussions reflect operational realities rather than theoretical assumptions. Progress toward interoperable governance approaches that enable safe cross‑border operation of public‑interest systems without requiring uniform technical architectures or centralized control models. A shared understanding that human oversight is contextual. In operational systems, effective oversight is often achieved through governance design, escalation protocols, auditability, and the ability to pause or shut down systems when conditions change, rather than continuous human‑in‑the‑loop interaction. This includes AI functioning as temporary digital coordination infrastructure, designed to operate only under predefined conditions and decommissioned once the public‑interest task is complete. Ultimately, success would mean that governance frameworks are informed not only by technical standards and legal principles, but by the lived experience of practitioners responsible for safe and accountable system operation.

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

These priority areas reflect challenges encountered in managing and governing digital systems that operate at scale, under regulatory scrutiny, and within public-interest environments. Safe, secure and trustworthy AI is critical when systems interact with large populations and support time-sensitive operations. Trust in these contexts depends not only on technical safeguards, but on disciplined scope definition, predictable behaviour, and clear responsibility structures. AI capacity-building is essential for public authorities and system operators who are increasingly expected to oversee AI-enabled services without the resources or technical depth of large technology vendors. Effective capacity-building must therefore emphasize governance literacy, escalation management, and risk awareness alongside technical knowledge. Interoperability of governance approaches matters because public-interest systems frequently span jurisdictions, organizations, and regulatory regimes. Governance models must allow systems to operate safely across these boundaries without forcing fragmentation or conflicting compliance obligations. Transparency, accountability, and human oversight ensure AI systems remain understandable and controllable. In many operational settings, continuous human-in-the-loop oversight is not feasible. Instead, oversight is most effective when designed into governance structures through auditability, predefined escalation pathways, and the ability to suspend or terminate system activity if conditions change.

A cross-cutting issue that deserves greater attention is the governance of situational and time-bound AI systems, systems intentionally designed to operate for limited durations and specific conditions, such as major public events, transport coordination, or emergency response. Many existing AI governance discussions implicitly assume persistent, continuously available systems. However, public-interest deployments often require AI that is deliberately temporary, narrowly scoped, and fully decommissioned once its operational purpose has been fulfilled. Governance frameworks should explicitly address how such systems are: Approved prior to deployment Monitored during operation Safely suspended or shut down Audited after completion A key challenge is how to proportionately govern AI systems whose risk profile may change rapidly during live operation, without subjecting temporary systems to the same approval, reporting, and audit burdens as long-lived deployments. Clarifying this lifecycle would enable responsible innovation while preventing unnecessary expansion of AI into domains where existing human or institutional processes remain more appropriate.

Across Europe and in public‑interest operational sectors, developments in AI governance are creating both important opportunities and practical challenges. On the opportunity side, emerging governance frameworks are helping move AI from experimentation toward trusted, mission‑specific deployment. Clearer expectations around safety, accountability, and oversight are enabling organizations to design AI systems with defined scope, explicit responsibility, and predictable behaviour. This is particularly valuable in public‑interest environments, where trust and reliability matter more than rapid innovation or scale. At the same time, governance gaps remain. Many existing frameworks implicitly assume persistent, continuously learning AI systems, which does not always align with how AI is used in operational settings. In sectors such as mobility coordination, large‑scale events, and public services, AI systems are often situational and time‑bound, operating under tight constraints and for limited durations. Applying governance requirements designed for long‑lived systems can create disproportionate compliance burdens, discouraging responsible use while doing little to reduce risk. Another challenge is uneven governance capacity. Public authorities and operators are increasingly expected to oversee AI‑enabled systems, but often lack the institutional experience, practical tooling, or shared language needed to do so confidently. This can lead either to over‑reliance on vendors or to overly restrictive interpretations of governance requirements, both of which carry risks. The key opportunity lies in developing governance approaches that are proportionate, interoperable, and lifecycle‑aware — enabling AI systems to be approved, monitored, paused, and decommissioned appropriately based on context. Doing so would support safe and responsible innovation while ensuring AI remains a controlled component of public‑interest infrastructure rather than an unmanaged expansion of automation.

The Global Dialogue on AI Governance can play a critical role by serving as a trusted coordination space where diverse governance approaches can be compared, understood, and aligned without forcing uniformity. Unlike bilateral or regional initiatives, the Dialogue is uniquely positioned to bring together governments and stakeholders with very different levels of technological maturity, regulatory capacity, and operational experience. Its value lies not in producing new rules, but in facilitating shared understanding of governance principles that can be adapted across contexts. The AI Dialogue can advance international cooperation by: Creating a common vocabulary around AI governance that bridges technical, operational, and policy perspectives. Providing a neutral forum to exchange practical lessons from real deployments, including successes, limitations, and governance challenges encountered in different sectors and regions. Supporting interoperability of governance approaches, enabling AI systems used in public‑interest contexts to operate safely across borders despite differing legal or institutional frameworks. Importantly, the Dialogue can reinforce the idea that international cooperation on AI does not require a single technical model or centralized control. Instead, it can promote mutual recognition of governance safeguards, lifecycle accountability, and proportional risk management. By anchoring cooperation in governance design and operational reality rather than technological competition, the AI Dialogue can help build trust between countries while supporting responsible innovation. This approach allows nations to retain sovereignty over their AI strategies while still participating in a shared global framework for safety, accountability, and public interest.

The AI Dialogue should build upon existing international and multistakeholder initiatives that have already contributed important foundations for AI governance, while providing added value by connecting them in a coherent and inclusive way. Relevant initiatives include: The Global Digital Compact, which establishes shared principles around an open, safe, and inclusive digital future. Work by organizations such as UNDP, ITU, UNESCO, and the OECD, which have developed guidance on AI ethics, capacity‑building, and policy coherence. Regional governance efforts, including emerging AI regulatory frameworks, standards bodies, and sector‑specific best practices. While these initiatives address important aspects of AI governance, they often operate in parallel, with limited connection to operational deployment experience across different sectors and regions. The added value of the AI Dialogue lies in its ability to: Act as a bridge between policy frameworks and operational practice, ensuring that governance discussions are informed by how AI systems are actually deployed, managed, and constrained in real‑world settings. Provide a continuous, inclusive forum where governments and stakeholders can compare governance approaches without duplicating existing mandates. Elevate attention to governance lifecycle issues, such as approval, oversight, suspension, and decommissioning of AI systems, particularly in public‑interest and civic contexts. By connecting existing initiatives through a practical, experience‑based lens, the AI Dialogue can reduce fragmentation, support learning across regions, and help translate high‑level principles into governance approaches that are usable, proportionate, and trusted in everyday practice.

Different stakeholders can contribute to the AI Dialogue most effectively when their participation is clearly structured around their distinct roles and forms of accountability. Governments can contribute by sharing policy priorities, regulatory experiences, and perspectives on sovereignty and public interest. The private sector can provide evidence from real‑world deployment, including how AI systems are scoped, governed, and constrained in practice. Academia and the technical community can support the Dialogue with research insights, methodological clarity, and evaluation tools. Civil society can contribute by highlighting lived impacts, inclusion considerations, and trust implications. To support this diversity of input, the AI Dialogue would benefit from a modular structure, including: Plenary sessions focused on shared principles and cross‑cutting challenges Thematic working sessions bringing together comparable system types or governance issues Practitioner‑level case discussions that focus on governance decisions rather than technical performance Maintaining a clear distinction between deliberation, knowledge‑sharing, and synthesis would help ensure discussions remain inclusive and productive. Structured listening sessions and curated written inputs can allow stakeholders with less capacity for continuous participation to contribute meaningfully. A format that balances openness with focused, purpose‑driven sessions would enable the AI Dialogue to reflect diverse perspectives while maintaining clarity and momentum.

Several perspectives remain underrepresented in global discussions on AI governance, despite being directly affected by AI deployment decisions. These include operational practitioners responsible for day‑to‑day system oversight, escalation, and accountability in public‑interest environments. Their perspectives are often absent from strategic discussions, even though governance decisions ultimately manifest through their responsibilities. Local and municipal actors are another underrepresented group. Cities and regional authorities frequently engage with AI in mobility, safety, service delivery, and event management, yet their experiences are not always reflected in national or global frameworks. Additionally, perspectives from temporary or situational AI deployments — such as systems used during large events, emergencies, or time‑limited coordination tasks — are often missing, as governance debates tend to focus on persistent or continuously deployed systems. Inclusion of these voices could be improved by: Creating dedicated practitioner or local‑authority engagement tracks Accepting concise written case inputs from operators and system owners Supporting participation by individuals and organizations without extensive policy or advocacy capacity Broadening participation in this way would help ensure AI governance reflects how systems are actually used, managed, and constrained in practice, not only how they are designed or regulated in theory.

Innovative engagement formats can help move the AI Dialogue beyond abstract discussion toward shared understanding grounded in practice. One effective format would be case‑based governance reviews, where participants examine anonymized or high‑level scenarios illustrating governance choices, trade‑offs, and escalation decisions rather than technical system details. This approach encourages constructive dialogue while avoiding sensitive or proprietary disclosures. Scenario‑based discussions could also be valuable, particularly for exploring how governance frameworks respond to changing conditions in real time, such as during public events, emergencies, or system failures. Short, structured "listening sessions" would allow underrepresented stakeholders to share experiences without needing to engage in open debate, helping ensure a wider range of perspectives informs the Dialogue. Hybrid formats combining in‑person sessions with asynchronous written submissions would further enhance inclusivity, enabling participation from stakeholders who cannot attend meetings or engage continuously. Finally, maintaining a clear focus on governance decisions and lifecycle management, rather than technical performance or product capabilities, would help keep discussions accessible, relevant, and anchored in the public interest.

Effective AI governance is most successfully advanced through practical, proportionate approaches that embed accountability into system design and operation, rather than relying solely on high-level principles or post-hoc controls. One promising practice is the adoption of lifecycle-based governance, where AI systems are approved, monitored, and evaluated according to their intended purpose, duration, and risk profile. This approach is particularly effective for public-interest applications, as it ensures responsibility is defined not only at deployment but also at suspension and decommissioning. Clear role and escalation frameworks are another important practice. Defining who is accountable for oversight, when escalation is required, and under what conditions systems can be paused or shut down enables human control without requiring continuous human-in-the-loop interaction. This has proven valuable in operational environments where time pressure and scale make constant intervention impractical. Capacity-building initiatives that focus on governance literacy rather than technical specialization also offer strong potential. Training public-sector operators and system owners to understand scope limitation, risk thresholds, and auditability helps reduce over-reliance on vendors and improves institutional confidence in oversight. At a policy level, interoperable governance principles, such as risk-based classification, transparency requirements proportional to impact, and mutual recognition of safeguards, support cross-border cooperation while respecting national and local contexts. Finally, structured use of case-based governance reviews and post-deployment audits provides a concrete mechanism for learning from real implementations. These practices allow governance frameworks to evolve based on operational experience, helping ensure AI remains a controlled and accountable component of public-interest infrastructure.