Department of Foreign Affairs and Trade, Ireland

Without responsible governance, new and emerging technologies have the potential to amplify disinformation; deepen digital divides; erode political freedoms; and pose significant security risks. Global cooperation in alignment with the Global Digital Compact will allow us to ensure that AI is developed in the service of humanity, and not to its detriment. The first Global Dialogue on AI Governance, as an outcome of the GDC, is an important opportunity to provide an inclusive space for governments and stakeholders to discuss responsible AI governance. A successful first Global Dialogue would facilitate greater cooperation and shared understanding of common challenges and approaches to AI governance, between governments, builders of technology, and civil society groups and independent experts. As per its aim to provide a universal home for AI governance cooperation, a successful Dialogue should ensure that the voices and priorities of less technologically advanced nations are reflected. Furthermore, the Dialogue should acknowledge the key role that civil society stakeholders play in discussions around AI governance, particularly in ensuring inclusive development and accountability. In terms of potential outcomes, the Dialogue should establish a clear follow-up mechanism that tracks implementation of agreed principles, ensuring that commitments translate into measurable accountability at national level. Additionally, the Dialogue could explore a roadmap overview of how this forum can enable a coherent and participatory approach to global governance, including how the Dialogue contributes to, supports, and fosters interoperability between existing forums and mechanisms such as the IGF.

• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

All of the thematic areas identified by General Assembly Resolution 79/325 are important and interrelated. A key priority for the UN Global Dialogue should be to establish clear baselines for the operation of AI within the context of the international human rights framework. The Dialogue is an important step in the implementation of the Global Digital Compact, and has the opportunity to advance progress in closing digital divides and promoting a safe, secure, trustworthy and human-centric approach to AI. Moreover, the Dialogue should be a space for surfacing concerns and enabling coordination of stakeholders on the socio-economic impacts of AI, which includes impacts on the furthest behind in terms of development as well as on employment and livelihoods, on technology sharing and infrastructure development, and environmental concerns. • Social, economic, ethical, cultural, linguistic and technical implications of AI: o AI has the capacity to manipulate narratives, spread and amplify misinformation and disinformation at speed, and undermine information integrity and diversity in information landscapes. This has implications which include the reinforcement of discrimination and inequalities. o As per the Council of Europe Guidelines on the Implications of Generative AI, the over-representation of the English language in LLMs has potential implications on the Freedom of Expression for users speaking less- and low-resourced languages. Due consideration should be given on how to leverage AI models to support less- and low-resourced languages. o A broader scope of responsible AI and AI safety could be considered, which incorporates the socio-economic implications of widespread AI production and use, including its impacts on labour, socio-economic inclusion, and agency over technology. • Protection and promotion of human rights: o It is vital that a human-centric and human-rights based approach is taken to the development of AI technology, from the design stage onward, in order to protect and uphold our freedoms and democracies in the digital age. • Transparency, accountability, and human oversight: o The rapid integration of AI as a content moderation tool, particularly without human oversight, raises concerns over moderation and the erosion of individual freedoms through intrusive surveillance, such as the freedom of expression, freedom of assembly, and freedom of association. o AI must be developed and used with trust, transparency and accountability in mind, particularly if it is to be deployed in public services. Impact assessments are also an important tool to identify and mitigate any potential adverse risks. o Accountability is essential for protecting people from potential adverse effects of AI, to ensure the safety of those who are most vulnerable and thus most likely to experience such negative effects. o Civil society voices have an important role to play in ensuring accountability at the local, national, regional, and international levels. • Interoperability of governance approaches: o Interoperability is a cornerstone of AI governance, with common terms and definitions necessary to form the foundation for global cooperation and interoperability across jurisdictions. o A focus on the interoperability of AI governance approaches is important to address the issue of fragmentation in the global governance landscape, which leads to conflicting guidance and rules, increased compliance burdens, and problems for innovation, safety, and equitable access to AI technologies. o The Dialogue can facilitate efforts towards shared governance language, incorporating transparency, accountability and oversight, and advance progress towards interoperable and compatible AI governance approaches.

The themes above do not fully capture concerns around democratic participation and value sharing in the development and use of AI systems. The production and control over AI systems is currently highly concentrated, which affects the autonomy and equity of states and other communities in developing contextually suitable and appropriate AI systems as knowledge infrastructure. The themes should engage with participation and agency in AI governance, apart from interoperability of current governance approaches. In particular, it is recommended that a specific thematic focus be incorporated which speaks to concerns around agency and equity in the development, use, and maintenance of AI across the supply and value chain. This track should cover labour conditions in AI data work, environmental sustainability, responsible minerals governance, infrastructure financing for the Global South, and data governance frameworks that address extractive cross-border data flows.

AI is a borderless technology and any effective solution to the challenges it poses cannot be addressed by a single country or bloc of countries. The Global Dialogue can serve as a multistakeholder platform to discuss international cooperation, share best practices and lessons learned, and facilitate open, transparent, and inclusive discussions on AI governance. Current global AI governance discussions could be perceived to suffer from a "governance divide", where norm-setting on AI safety, standards and IP could be seen to reflect developed-country priorities, with issues central to the Global South (labour conditions in AI production, equitable value chains, critical minerals, language model performance disparities) viewed as marginalised. The Dialogue is an important opportunity to address this issue by bringing all Member States and stakeholders to the table to discuss the governance challenges that cannot be solved without meaningful global participation. The Dialogue is vital in establishing a politically legitimate, multilateral and multistakeholder mechanism for global cooperation on AI. As per existing analysis, as of 2024, 118 countries were not party to any significant international AI governance framework, and the Dialogue's universal membership basis and institutionalisation within the UN system will be an important step towards addressing this gap. The Dialogue must also address the issue that important features of AI governance today are impacted by the internal policies and rules of private technology companies and AI developers. Even as regulatory standards are being developed by countries and regions, the Dialogue should provide a mechanism for discussing, monitoring and evaluating corporate compliance with global standards, including standards like the UN Guiding Principles on Business and Human Rights. In many respects, the principles informing the standards are not new; it is the area of applicability and the consequences of the impact which are.

While there have been substantial efforts to develop AI global governance and cooperation frameworks already and progress has been made, a fragmented governance landscape remains a risk that the Dialogue can seek to address with its inclusive, multistakeholder model and universal mandate, as well as its potential to bring coherence in an interoperable and compatible way to existing international initiatives on AI. The Dialogue should leverage, for example, existing initiatives and processes within the UN system. These include, pertinently, the IGF as an inclusive and multi-stakeholder forum for questions of internet governance, as well as processes, frameworks and principles developed by the OHCHR, UNESCO, UNCTAD and ITU, among other sector-specific work by agencies like UN Women, UNDP and UNEP. The Dialogue can serve as a bridge and a mechanism to translate the important principles and frameworks developed by bodies like UNESCO into actionable commitments for members. The Dialogue could also examine synergies with other international organisations and initiatives including the EU Independent Scientific Panel on AI and the EU AI Act Advisory Forum, the G7, G20, the OECD AI Policy Observatory, the Global Partnership on AI (GPAI), and the AI Summit series, with the aim of connecting efforts, building common ground on interoperability and shared approaches. Finally, as identified by the Co-Chairs, the Dialogue must work closely with the Independent Scientific Panel, not only to leverage its independent expertise towards informing AI governance discussions, but to ensure that the panel takes into considerations the real-world social, economic and governance dimensions that the Dialogue surfaces.

The value of the AI Dialogue lies in its ability to move beyond traditional UN plenary sessions, and to function as a genuine multistakeholder and inclusive platform. The Dialogue should be organised with input from all stakeholders in mind, ensuring cross-stakeholder mix in participation and active roles for all participants. Participatory and interactive sessions would facilitate open, transparent and inclusive dialogue, with a focus on practical exchange of ideas and experiences, and allow for the generation of concrete proposals and formation of new partnerships. To ensure the broadest range of perspectives are heard, private sector representation should be broad and inclusive. Along with the upcoming in-person Dialogues in Geneva in 2026 and New York in 2027, the Dialogue could also consider incorporating regular, thematic consultations organised around the Dialogue's core themes, which would feed into the in-person engagements. Consultations which are open to all Member States and stakeholders would contribute to ensuring ongoing, meaningful participation, building on the momentum from July 2026 towards May 2027. Ideally, participation, engagement, and accreditation should take place through a bottom-up process (such as in the IGF). Additionally, the Global Dialogue track should establish a clear interface and dialogue with the IGF and existing multi-stakeholder initiatives and mechanisms under the UN auspices, like the WSIS. The recommendations of the Netmundial + 10 outcome document provide useful guidance in this regard (https://netmundial.br/pdf/NETmundial10-MultistakeholderStatement-2024.pdf) .

The potential benefits of AI are not universal, and countries with less developed AI infrastructure often have less influence over how AI is governed globally, even when the impacts on their populations are significant. An important priority, as outlined in the EU's International Digital Strategy, is to further include voices and perspectives from the Global South in these conversations. Ensuring that the mandate of the Global Dialogue to be an inclusive forum is fulfilled requires commitments to help Member States, and other relevant stakeholders, particularly low-resourced states, develop the capacity for structured participation in the Global Dialogue. It is welcome that support for participation, particularly for developing country delegations, is being actively pursued through existing UN system mechanisms and voluntary contributions.

As the AI Dialogue is not a traditional UN format, in order to be successful, it should take a non-traditional approach. It will be important to ensure that the Dialogue takes an interactive and dynamic approach in order to maximise engagement. Therefore there should be an appropriate balance between allowing adequate time for plenary segments and adequate space for important multistakeholder sessions. This will allow for a maximum of perspectives to be shared, and facilitate a more participatory dialogue. These perspectives should come from those who can speak to the impacts of AI, as well of those who understand its potential, ensuring that the Dialogue facilitates a balanced and informed discussion. The Dialogue could explore the use of break-out sessions, panels featuring a wide range of voices, alongside the use of interactive technology.

EU Governance The EU AI Act is the European AI governance framework for the whole European market. It follows a risk-based approach, introducing rules for the development and use of AI systems that are commensurate to the risks, and do not go beyond what is necessary. The AI Board gathers all EU Member States around the consistent and effective implementation of the legislation at national level and the European AI Office leads on implementation of the AI Act, fostering AI innovation and excellence in AI and positioning the EU as a global leader on AI governance. The EU Independent Scientific Panel is being appointed to support the scientific understanding of AI capabilities and the AI Advisory Forum will complement this work, serving as a general advisory body to the European Commission. The AI Pact is a multistakeholder platform for companies and organisations to prepare for the different stages of implementation of the AI Act. The Pact gathers and allows for the network to exchange best practices and experiences while learning about the EU approach to AI. This initiative also encourages organisations to proactively disclose the processes and practices they are implementing to anticipate compliance. Labour Governance Partnership on AI Responsible Sourcing Guidelines for Data Enrichment Services - standards for labour conditions in data annotation and content moderation work (https://partnershiponai.org/wp-content/uploads/2022/11/data-enrichment-guidelines.pdf) Fairwork Institute - certification and benchmarking of decent work standards for platform and data workers (https://fair.work/en/fw/certification/) Labour20 G20 Statement on Decent Work in the Platform Economy - call for a binding ILO Convention on platform economy labour protections (https://tradeunionsinafcfta.org/wp-content/uploads/2025/11/L20-statement-to-the-G20-labour-and-employment-ministers_2025.pdf) Minerals Governance OECD Due Diligence Guidance for Responsible Mineral Supply Chains - responsible sourcing framework for critical minerals including those used in AI hardware (https://www.oecd.org/en/topics/policy-issues/responsible-business-conduct.html) Responsible Minerals Initiative - environmental and labour standards in mining industries linked to AI infrastructure (https://www.responsiblemineralsinitiative.org/) Environmental Sustainability EU Energy Efficiency Directive - mandatory energy performance and sustainability reporting for data centres (https://energy.ec.europa.eu/topics/energy-efficiency/energy-efficiency-targets-directive-and-rules/energy-efficiency-directive_en) UNEP Sustainable Procurement Guidelines for Data Centres and Servers - sustainability standards for public procurement of AI infrastructure (https://www.unep.org/resources/toolkits-manuals-and-guides/sustainable-procurement-guidelines-data-centres-and-servers) UNEP Draft Resolution on Environmental Sustainability of AI Systems - emerging multilateral instrument for environmental governance of AI (https://docs.un.org/en/UNEP/EA.7/L.14) AI Infrastructure and Compute Access EuroHPC Joint Undertaking - EU model for regionally governed public compute infrastructure, cited as a blueprint for Global South adaptation (https://www.eurohpc-ju.europa.eu/index) Masakhane Languages Hub - pan-African initiative for African language representation in AI training data and model development (https://www.masakhane.io/home) Public AI White Paper - Bertelsmann Stiftung - framework for public compute as an alternative to privatised AI infrastructure (https://www.bertelsmann-stiftung.de/de/publikationen/publikation/did/public-ai-white-paper-a-public-alternative-to-private-ai-dominance) Open Source and Decentralised AI Development Masakhane InkubaLM - small language model developed for African languages outside major AI lab control LatamGPT - Latin American initiative for regionally developed Spanish- and Portuguese-language models (https://www.latamgpt.org/en)