Siemens Energy

In my view, the first Global Dialogue on AI Governance will be successful if it moves beyond high-level principles and delivers concrete, operational, and inclusive outcomes. A key priority is to establish a shared baseline of definitions and risk taxonomies. Today, differences in how concepts such as "high-risk AI" or "frontier models" are defined create fragmentation and limit effective coordination across jurisdictions. I also see strong value in developing a clear roadmap for interoperability of governance approaches. With multiple regulatory frameworks emerging globally, identifying practical alignment mechanisms, such as mutual recognition or minimum safeguards, will be critical. Another important outcome would be tangible commitments on AI capacity-building, particularly for countries and institutions that currently lack the technical and regulatory capabilities to engage fully in AI governance. Without this, global discussions risk remaining uneven and exclusionary. In addition, I believe the Dialogue should lead to the creation of multi-stakeholder implementation coalitions focused on priority areas such as AI safety, evaluation, and public-sector use. These should be action-oriented, with clear deliverables.

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

These priorities reflect where I see the most urgent need for practical progress in global AI governance. For me, safe, secure and trustworthy AI is foundational, especially given the rapid advancement of general-purpose and increasingly autonomous systems. There is a clear need for more alignment on risk assessment, safety evaluation, and monitoring practices. I consider AI capacity-building to be essential for ensuring that AI governance is globally inclusive. Many countries and institutions still lack the resources and expertise needed to participate effectively, which risks reinforcing existing inequalities. Interoperability of governance approaches is another key priority. As regulatory frameworks continue to evolve across regions, fragmentation is becoming a real challenge. I believe there is a strong need for practical mechanisms that enable alignment while respecting different legal and cultural contexts. Finally, transparency, accountability, and human oversight are critical to making governance frameworks effective in practice. This includes improving standards for documentation, auditability, and oversight of AI systems, particularly in high-impact use cases.

In addition to the listed themes, I see several important cross-cutting and emerging issues that deserve more attention. One is the rise of agentic AI systems, which pose new challenges regarding autonomy, decision-making, and control. These systems go beyond traditional models and require updated governance approaches. Another important issue is computer governance and the concentration of power. Access to large-scale compute and data is becoming a key determinant of who can develop advanced AI systems, raising concerns about inequality and market concentration. I also see a gap in evaluation and auditing standards for advanced AI systems. There is a need for more robust, internationally recognized approaches to testing, benchmarking, and independent auditing, especially for high-capability models. In addition, public sector use of AI deserves more focused attention. Governments play a dual role as both regulators and users of AI, creating specific governance challenges, particularly in areas such as procurement and accountability.

From my perspective, working at the intersection of AI, policy, and implementation in Europe, governance gaps in the selected areas are already having tangible impacts on both the public and private sectors. One of the most significant challenges is the fragmentation of governance approaches, even within relatively aligned regions such as the European Union. While frameworks like the EU AI Act provide a strong foundation, differences in national implementation, sectoral interpretation, and alignment with non-EU jurisdictions create uncertainty for organizations operating across borders. This increases compliance costs and slows down innovation cycles. Another key gap lies in operationalizing trustworthy AI. Many organizations struggle to translate high-level principles into concrete processes such as risk classification, model evaluation, and lifecycle monitoring. This is particularly evident with the rapid emergence of generative and agentic AI systems, where existing governance tools are often insufficient. Capacity constraints also remain a critical issue. While Europe has strong regulatory ambition, there is a shortage of technical expertise within public institutions to effectively audit, procure, and oversee AI systems. This creates an imbalance between regulatory expectations and implementation capabilities. At the same time, these challenges create important opportunities. The EU is well-positioned to lead in setting global standards for trustworthy and human-centric AI, particularly through its regulatory frameworks and policy leadership. There is also growing momentum around AI governance tooling, auditing services, and compliance innovation, which opens new avenues for both public-private collaboration and market development.

In my view, the AI Dialogue can serve as a neutral platform to move from fragmented discussions toward real coordination and joint action. It can help establish a shared global baseline by aligning terminology, risk classifications, and core governance principles across jurisdictions, reducing unnecessary complexity without enforcing a one-size-fits-all approach. I also see strong value in its role as a coordination layer between existing initiatives, including efforts led by the United Nations and the OECD. Connecting these efforts can improve coherence and avoid duplication. In addition, the Dialogue can catalyze multi-stakeholder, action-oriented coalitions in areas such as safety evaluation, auditing, and capacity-building, with clear deliverables and timelines.

In my view, the AI Dialogue should build on and connect existing initiatives rather than duplicate them, with a focus on coordination and interoperability. Key efforts include the OECD AI Principles and policy observatory, the Global Partnership on AI (GPAI), and UNESCO's UNESCO Recommendation on the Ethics of AI. In addition, regulatory developments such as the EU AI Act and national AI strategies provide important implementation experience. Technical standard-setting bodies like ISO and IEEE also play a critical role in operationalizing governance. However, these initiatives often operate in parallel, with limited coordination and varying levels of global inclusiveness. The added value of the AI Dialogue lies in its ability to act as a bridging and orchestration platform. It can connect policy, technical, and implementation communities, enabling alignment between high-level principles, regulatory frameworks, and standards.

In my view, the effectiveness of the AI Dialogue will depend on structured, role-based contributions from different stakeholder groups, combined with a format that enables both inclusivity and concrete outcomes. Governments should contribute by sharing regulatory experiences, policy frameworks, and implementation challenges, including lessons learned from instruments such as the EU AI Act. This is critical for identifying convergence points and gaps. Private sector actors can provide technical expertise, real-world deployment insights, and emerging risk scenarios, particularly for advanced AI systems. Their input is essential for ensuring that governance approaches remain practical and future-proof. Academia and research institutions should contribute evidence-based analysis, evaluation methodologies, and foresight on emerging risks, while civil society organizations play a key role in representing societal impacts, human rights considerations, and inclusion perspectives. In terms of format, I recommend a multi-layered structure. This could include: 1. High-level plenary sessions to align on strategic priorities 3. Thematic working groups focused on specific areas (e.g., safety, interoperability, capacity-building) 3. Action-oriented task forces with clear mandates, deliverables, and timelines To ensure continuity, the Dialogue should establish ongoing working mechanisms, rather than a one-off event. This includes regular follow-ups, progress tracking, and open knowledge-sharing platforms.

Currently global discussions on AI governance still lack balanced representation across regions, sectors, and lived experiences. First of all, stakeholders from the Global South remain underrepresented, particularly policymakers, technical experts, and local innovators. This limits the relevance of governance frameworks in contexts where infrastructure, labor markets, and data ecosystems differ significantly. Inclusion requires targeted capacity-building, funding for participation, and regional consultation mechanisms embedded into global processes. Also, public sector practitioners, especially from municipalities and service delivery agencies, are often missing from high-level discussions. Yet they are among the primary deployers of AI in areas such as welfare, healthcare, and education. Their inclusion would strengthen the practicality of governance approaches. At the same time, small and medium-sized enterprises and startups are underrepresented compared to large technology companies. Their constraints and innovation dynamics differ significantly, and governance frameworks should better reflect this. Mechanisms such as dedicated SME tracks or advisory panels could help address this gap. Also, there is still limited representation of civil society actors from diverse cultural and linguistic backgrounds, as well as communities directly affected by AI systems. This includes workers impacted by automation, marginalized groups exposed to algorithmic bias, and non-English-speaking communities. Inclusion here requires localized engagement formats, multilingual access, and participatory mechanisms beyond formal consultations. To address these gaps, the AI Dialogue should adopt a deliberately inclusive design, combining funding support, regional outreach, hybrid participation formats, and structured roles for underrepresented groups in decision-shaping processes, not just consultation.

I would prioritize interactive, outcome-oriented formats over traditional panels. First, I would suggest policy labs where small, mixed groups co-develop concrete outputs (e.g., risk frameworks, audit templates) based on real use cases. Also, it might be beneficial to have scenario simulations to test governance approaches against frontier and agentic AI risks, helping stakeholders understand gaps in current frameworks. Implementation sprints (1–2 days) focused on delivering specific artifacts, such as evaluation benchmarks or documentation standards, are also commonly beneficial for this kind of dialogue.

EU AI Act: risk-based framework with concrete obligations (risk classification, conformity assessments, requirements for high-risk systems) OECD AI Principles: widely adopted baseline for trustworthy AI UNESCO Recommendation on AI Ethics: strong focus on human rights and societal impact Algorithmic Impact Assessments (AIAs): a practical tool for pre-deployment risk evaluation, especially in the public sector Model cards/system cards / AI audits: improving transparency, documentation, and accountability in practice ISO & IEEE standards: translating principles into measurable technical requirements Regulatory sandboxes: enabling controlled experimentation and regulator-innovator collaboration