Switzerland

The first Global Dialogue on AI Governance should ensure a full multistakeholder approach, enabling the meaningful participation of all stakeholder groups on an equal footing taking into account their respective roles and responsibilities. Switzerland believes the inaugural Dialogue should aim to produce an ambitious Chair's summary identifying key governance challenges, areas of convergence, and possible avenues for future cooperation. This can be achieved by highlighting existing initiatives and partnerships, thereby harnessing synergies across institutions and stakeholder communities. Through transparency, trust-building and open dialogue, the process can advance international cooperation on AI governance. The outputs of the Dialogue should contain actionable policy outcomes. This includes fostering agreed principles and political commitments that can serve as a basis for regulatory development across different national contexts. The Dialogue should also function as an agenda-setting platform, helping to bring together currently fragmented processes. In this regard, cooperation with organizations such as the ITU, UNESCO, the UN IGF and other relevant UN agencies, particularly in the Geneva context, can strengthen synergies and consolidate existing work on AI governance. By integrating contributions from the Human Rights Council and the Office of the High Commissioner for Human Rights, the Dialogue can ensure that outcomes remain inclusive and grounded in a human rights-based approach, thereby protecting and promoting fundamental values.

• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Open-source software, open data and open AI models

- Firstly, the Global Dialogue is key in fostering coherence and a common understanding of, as mentioned in paragraph 1 of the WSIS declaration of principles, people-centred, inclusive and development orientedAI governance. These priorities therefore reflect the need to anchor the UN AI Dialogue in an inclusive, practical and implementation-oriented approach. - In the current geopolitical and economic context, prioritising open software, open data and open AI models is essential. AI systems should be accessible to all actors, including those with limited resources, and enable countries and organisations to develop and appropriate their own systems, as reflected in initiatives such as Apertus and the framing of AI as part of digital public infrastructure. This also strengthens technological autonomy and reduces dependency on a limited number of providers. In this context, ensuring portability, meaning the ability to switch providers and transfer data across systems, is a key requirement to maintain flexibility, foster competition and support a more inclusive and resilient AI ecosystem. - Strengthening AI capacity-building is essential to ensure that participation is meaningful, as not all governments and non-governmental actors, particularly in the Global South, have equal resources to engage; empowering civil society actors to contribute brings perspectives grounded in lived realities. - Promoting interoperability of governance approaches further enables the harmonisation of frameworks while supporting open innovation and open-source ecosystems, thereby facilitating practical cooperation across stakeholder communities. In this regard, interoperability of governance approaches can be seen as the counterpart to portability between AI systems. In this context, the dialogue's added value lies in providing a neutral global space to exchange best practices, identify governance gaps and foster collaboration. (use the existing IGF, WSIS Forum, CSTD create synergies...) - Finally, transparency, accountability and human oversight are critical to enable meaningful discussions and to build trust in both the process and its outcomes. In this context, ongoing work on technical standards, particularly within ISO and ITU, can be closely integrated, as these provide concrete, operational frameworks that can support the implementation of agreed principles.

The environmental impact of AI could be captured more explicitly and can be reflected across the discussions. This includes not only the energy and water consumption associated with the development, training and deployment of AI systems, but also the broader sustainabilitiy implications of digital infrastructures. In addition, the meaningful participation of underrepresented and vulnerable groups should be treated as a cruss-cutting priority. AI governance processes should focus on enabling the effective participation of these groups, including youth and women, as this is essential for both the legitimacy of the process and for the quality of outcomes as integrated perspectives of vulnerable groups can help better identify risks, needs, and opportunities that may otherwise be overlooked. Here, it is also important to underline that smaller and less well-resourced communities should have meaningful access to, and choice between, different AI systems. They should be able to select the systems that best meet their specific needs, appropriate and adapt them to their contexts, and further develop them over time. This is essential to ensure more equitable participation, reduce dependency on a limited number of providers, and support a more inclusive and balanced AI ecosystem. Finally, greater attention could be paid to the interplay between technical standards and AI policy. Standards and policy instruments risk being discussed in parallel, without taking into account their relationship throughout the process. Technical standards can help operationalize policy objectives, while policy can provide the direction for the standard-setting efforts. A more explicit discussion of how standards and policy can complement one another could strengthen the overall government approaches to AI governance.

Switzerland currently observes a capacity gap within the public sector, where many administrations are not yet equipped with the necessary knowledge and institutional capacity to keep pace with the rapid development of AI technologies. In parallel, the adoption and operational use of AI remain uneven across sectors, with areas such as the judiciary and healthcare facing greater challenges in implementation compared to others. Finally, international cooperation and interoperability remain constrained, as governance approaches are either insufficiently developed or primarily focused on national priorities. This can limit alignment across jurisdictions and reduce opportunities for more effective cross-border cooperation. The challenges currently faced are a set of interrelated challenges that begin with capacity constraints, particularly within the public sector, where institutions often lack the expertise and resources required to keep pace with the rapid development and deployment of AI technologies. At the same time, the adoption and effective use of AI remain uneven across sectors which can lead to asymmetries in implementation and preparedness. Such asymmetries are also visible in government approaches that are still currently fragmented. Diverging national strategies and regulatory frameworks can limit interoperabilitiy and hinder international cooperation which can strain the overall effectiveness of global AI governance efforts. Finally, financing remains a critical cross-cutting constraint. Adequate and sustainable funding is required to support capacity building, regulatory development, technical standard-setting and international coordination. However, these challenges also present significant opportunities. Through active best-practice sharing and multistakeholder cooperation, such as initiatives like the International Computation and AI Network (ICAIN), which aim to lower barriers, democratize access to AI capabilities, and foster sustainable innovation, the dialogue can help establish a more cooperative and inclusive approach to AI governance. In this context, the Dialogue has the potential to serve as a catalyst for building a network that strengthens coordination and addresses existing governance gaps. By leverging Geneva as a platform for these discussions, the concentration of expertise there, such as human rights and technical standard-setting can offer a solid foundation to foster synergies and enhance the coherence of international AI governance efforts, anchored in a human-centred approach and respect for international law.

The Global AI Dialogue can play a central role in advancing international cooperation on AI governance by providing a dedicated UN platform for open, inclusive, multilateral and multistakeholder exchange. It sets an important precedent for inclusive AI governance, grounded in human rights and international law, and creates a neutral global space to exchange best practices, identify governance gaps and foster practical cooperation across stakeholder communities. In this context, the Dialogue offers a unique opportunity to strengthen cooperation, promote interoperability between different governance approaches, and address capacity and implementation gaps, particularly for developing countries. It is further reinforced by concrete mandates stemming from the WSIS+20 Outcome Document, particularly the task assigned to the UN Inter-Agency Working Group on AI to map existing capacity-building efforts, identify gaps and report thereon within the Dialogue, thereby linking discussions to actionable follow-up, while also enhancing coherence across processes by building bridges with UN CSTD data governance discussions and connecting with the IGF and WSIS frameworks, further strengthening international cooperation. In addition, the possibility of organising consultations in diverse geographic locations beyond the Global North can enhance meaningful participation from stakeholders in the Global South, ensuring that international cooperation is informed by a broader range of perspectives, making international cooperation truly international. Through all these points, there is a clear vision to create a better synergetic basis for international cooperation by leverging all the previous and current work which bring the right actors into the same space that provides a space to create policy continuity instead of a standalone discussion that addresses what we want to deal with international and nationally. By using these mechanisms to work through all the complex layers of AI, the dialogue can create a common understanding for all actors to set the stepping stones for successful future work.

The AI Dialogue should build on and connect with existing multilateral, regional and technical frameworks to avoid duplication and maximise coherente outcomes. At the global level, the WSIS architecture provides a key foundation comprised, inter alia, by the United Nations Group on the Information Society (UNGIS), the Commission on Science and Technology for Development (CSTD), and the Internet Governance Forum (IGF), which already facilitate inclusive and multistakeholder discussions on digital governance. Furthermore, initiatives and frameworks developed by the Council of Europe through the Council of Europe Convention on AI, a global instrument for AI regulation, the OECD through the GPAI offer important standard and policy references, that can contribute to shaping common principles and approaches. In addition, Geneva-based human rights institutions play a central role in ensuring that AI governance remains grounded in international human rights law and standards. Finally, the UNESCO recommendation on Ethics of AI, the first global normative instrument setting a standard for ethical AI, to prioritize human rights, dignity, and environmental sustainability. Similarly, the ISO, IEC ITU and IEC standard-setting bodies can be built upon and connected with to translate principles into operational practices. Finally, capacity-building initiatives, such as those led by DiploFoundation and its Geneva Internet Platform, also provide valuable models for strengthening expertise and enabling more inclusive participation, particularly from underrepresented stakeholders.

By leveraging and actively using the São Paulo Multistakeholder Guidelines, which were recognized by paragraph 94 of the WSIS+20 Outcome Document, the Dialogue can ensure inclusive, transparent and balanced participation across all stakeholder groups, thereby strengthening the legitimacy, credibility and effectiveness of its outcomes. Switzerland also sees merit in structuring discussions along thematic clusters that build on existing international expertise and institutional experience within the UN system and other multilateral fora. In this regard, the following clusters could be considered: First, anchoring trust and safety by linking technical robustness with human rights and accountability (topics a, e, f), and leveraging Geneva's role as the home of the UN Human Rights Council; second, addressing inclusion and capacity to bridge the digital divide and manage socio-economic impacts (topics b, c) through existing UN mechanisms; and third, fostering interoperability and open innovation by harmonizing governance approaches while supporting open-source ecosystems (topics d, g). This should be done through meaningful participation ensured through an inclusive session design, including formats where government and high-level representatives engage in the same room alongside other stakeholders.

Underrepresented groups in AI governance, including people from developing countries, women and girls, youth, older persons, persons with disabilities, Indigenous peoples, migrants, refugees and internally displaced people and those in vulnerable situations, should be addressed sufficiently in the Global AI dialogue, consistent with the recommendations contained in the São Paulo Multistakeholder Guidelines, especially their "process steps". In particular, these guidelines emphasise inclusivity, transparency, accessibility and iterative participation across all stages of policy processes. To do so, mechanisms for inclusion of marginalized people and groups can consider: participation modalities, including flexible time zone arrangements, to ensure broad and equitable access. Stakeholders, particularly civil society and smaller actors, must be able to contribute continuously throughout the process, including through structured opportunities for written submissions, in line with the Guidelines' call for open, continuous and well-documented consultation processes. Building on lessons from the WSIS+20 process, it is essential to create dedicated spaces for meaningful multi-stakeholder engagement, especially with civil society, while maintaining a careful balance between substantive thematic discussions and high-level exchanges. To support informed participation, progress reports and summaries should be made available in a timely manner, enabling stakeholders to provide relevant written input at different stages. This requires a clearly defined timeline that outlines key phases of the process and submission windows, as encouraged by the São Paulo Guidelines' emphasis on predictability and clarity of process design.Finally, consideration should be given to diversifying the geographical footprint of the Dialogue by exploring locations beyond New York, thereby reinforcing inclusivity of underrepresented voices from around the world.

A format involving long plenary statements could hinder meaningful and dynamic engagement during the dialogue. Therefore, it would be helpful to instead adopt more interactive and varied formats. Past conferences could provide insights into useful formats. Furthermore, it is important to strike a healthy balance between formats that allow for Higher level visibility and enabling in-depth discussions on specific points so as not to disadvanatage smaller delegations. The involvement of non-governmental stakeholders is also essential to open the discussion to different perspectives. Breakout discussions and roundtables could also enable a deeper more focused exchange between smaller groups.

- The Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law is an international treaty on artificial intelligence adopted under the auspices of the Council of Europe. It constitutes a global instrument open to all countries, enabling them to ensure that the development and use of AI systems are aligned with human rights, democracy, and the rule of law, while also fostering innovation. Since its opening for signature on 5 September 2024, the convention has been signed by a growing number of countries from different regions of the world, reflecting its broad international relevance and applicability. - HUDERIA is the methodology More than 50 countries from all regions of the world contributed to the Huderia initiative, underscoring its broad international support and relevance. This wide participation is particularly important for advancing global interoperability of risk assessment mechanisms, as it helps promote shared approaches, reduce fragmentation and enable more consistent and coordinated implementation across jurisdictions. - The OECD AI Principles are the first intergovernmental standard on AI, promoting innovative, trustworthy AI that respects human rights and democratic values. With over 50 signatories, the principles were first adopted in 2019 and updated in 2024. The OECD's definition of an AI system, is widely considered the most accepted intergouvernmental definition, serving as a foundation for legislation in the EU, the US, and the G20. - The Sao Paolo Multistakeholder Guidelines, are a main outcome of the NetMundial+10 meeting, held in São Paulo from 29 to 30 April 2024, and translate the 2014 NETmundial Internet Governance Process Principles into practice. They provide concrete, non-binding recommendations to strengthen multistakeholder processes and ensure they are inclusive, transparent and participatory, not just in name but also in practice. They were recognized in paragraph 94 of the WSIS+20 Outcome Document adopted by the UN General Assembly. - The Internet Governance Forum (IGF) serves as the primary global platform for multi-stakeholder discussions on digital policy issues, and in recent years, artificial intelligence has emerged as one of its central and most actively addressed themes. - The UNESCO Recommendation on the Ethics of Artificial Intelligence, adopted in 2021, is the first global standard-setting instrument on AI ethics. It provides a common normative framework to guide the development and use of AI in a manner that respects human rights, human dignity, and fundamental ethical principles.