Centre for Communication Governance, National Law University Delhi (CCG-NLUD), on behalf of the 'Multistakeholder Approaches to Participation in AI Governance' (MAP-AI), an initiative launched by the Global Network Initiative (GNI) and the Centre for Communication Governance (CCG), with the objective to foster meaningful and effective multistakeholder engagement across a range of critical AI governance-focused convenings, processes, and initiatives, with a particular focus on elevating underrepresented voices and perspectives.

The UN Global Dialogue (UNGD) is well-positioned to: -Model meaningful multistakeholder engagement: Operationalise meaningful engagement for a wide range of stakeholders, especially civil society actors (defined broadly to include independent non-profits, academic institutions, researchers, and technical experts), who add relevant expertise, diverse perspectives, enhance accountability and legitimacy, and are excluded from AI governance processes. Specifically, it should highlight less-represented perspectives, including those from grassroots organizations and local communities of NLP researchers and developers, and integrate them into the outcomes of this process. -Contribute coordination and coherence across AI Governance processes: Find pathways for alignment, coherence, and cross-pollination across governance structures and processes. This includes advancing the outcomes from, and exploring opportunities to coordinate across key UN, global and regional AI/tech governance processes and bodies (e.g., the AI Summit series, OECD/GPAI, Hiroshima AI process, Internet Governance Forum (IGF), CSTD, WSIS Forum). E.g., the Independent International Scientific Panel could present its findings to the WSIS communities and vice-versa. -Emphasise financing and capacity building: Highlight and take forward proposals already underway within the UN and through the Global Digital Compact (GDC), importantly, the Global Fund on AI, and identify ways to develop appropriate financing mechanisms, and a roadmap for capacity-building initiatives for implementation at the country level. One instructive model could be the UNDP's capacity-building system, which combines core resources, government and third-party financing/cost-sharing, and UN pooled funds, and leverage existing UN bodies at the regional and national level. UNGD should create dedicated financial resources for AI capacity-building with governments and civic actors in the Global Majority. -Drive accountability: Facilitate the multistakeholder development of concrete commitments, based on international legal frameworks, by member states and the private sector that will contribute to rights-based, democratic AI governance, and create mechanisms to hold actors accountable to these commitments.

• Safe, secure and trustworthy AI
• AI capacity-building
• Interoperability of governance approaches
• Protection and promotion of human rights

We note the UNGD Co-Chairs' "Draft Note on themes and structure of the Global Dialogue", released subsequent to the publication of this submission form. We have framed our explanation in terms of the themes listed in GA/R/79/325, and support the Co-Chairs proposed thematic clusters. -Safe, secure and trustworthy AI: UNGD should build on the Independent Scientific Panel's work to establish a clear taxonomy of AI risks. Here, OHCHR's taxonomy of AI risks (2023) is instructive. UNGD should move beyond frontier risks like nuclear and biosecurity in its approach to "AI safety", and ensure that risks related to cybersecurity, human rights, and democracy are prioritized. It should emphasise lived experiences of vulnerable populations, particularly from the Global Majority, such as the exploitative and gendered implications for the data labeling workforce. Mechanisms can include shared frameworks for incident reporting, risk assessment (particularly, human rights impact assessments), and auditing. -AI capacity building: Please refer to response numbers 8 & 12. -Interoperability of governance approaches: The UNGD should build interoperability between AI governance frameworks by identifying shared objectives and outcomes, ensuring coherence in implementation, and tracking outcomes and performance against commitments/deliverables. This also enables countries and stakeholders to draw insights for aligning domestic governance mechanisms, while preserving policy space, democratic oversight, and digital sovereignty across different regional and national contexts. -Protection and promotion of human rights: It is critical to ground the UNGD processes, discussion, and outcomes in international human rights framework as it allows commitments and processes to be coherently articulated, and critically examined, while highlighting the risks present in contexts with stark societal inequalities and diminished institutional capacity. UNGA Resolution A/RES/79/325 identifies protection of human rights as a priority, which provides it with legitimacy to focus on key elements missing in other ongoing AI processes.

-Context-driven AI infrastructure: UNGD should promote commitments to support the development of context-driven AI infrastructures, focusing on structural enablers for equitable participation in AI ecosystems. This includes enabling the development of local AI ecosystems, capacity building, access to compute resources, availability of local and representative datasets, support for multilingual systems, and the involvement of local and indigenous communities, and national research institutions towards building AI that is responsive to diverse socio-economic contexts and grounded in the lived realities of communities. -Environmental Sustainability: AI systems, data centres, and semiconductor production depend on significant natural resources, placing pressure on ecosystems and vulnerable communities. A key challenge is that AI's environmental impacts remain difficult to isolate and thus hard to assess without transparent corporate disclosures. The UNGD should prioritise mandatory environmental reporting frameworks for AI companies, moving beyond narratives that frame AI only as a sustainability tool. This should also encourage alignment with existing initiatives on transparency, anti-corruption, and human rights due diligence in extractive and critical mineral supply chains connected to AI infrastructures. The UN Guiding Principles on Business and Human Rights offer a normative basis for such obligations. The UNESCO AI Ethics Recommendation provides further normative and practical support. -Data governance and data justice: Data governance is a foundational layer for effective and equitable AI governance. The UNGD should build on existing processes, such as the Data Governance Working Group established under the GDC, as well as commitments around data stewardship, access, interoperability, and the mitigation of structural inequalities embedded in data ecosystems. It is important to understand and address these inequalities from a gendered and intersectional lens. Other important cross-cutting topics include: AI in conflict contexts, workforce/labour issues in the AI value chain; and the gender implications of AI.

#NAME?

The UNGD can encourage universal participation by playing the following concrete roles- (1) Building common governance priorities and vocabulary: Across countries, there can be several shared priorities in AI, yet there is a lack of pathways for countries with different regulatory logics, institutional capacities, and development priorities to find workable common ground across these. A common governance vocabulary would mean, for instance, that when one jurisdiction classifies an AI system as "high-risk" and another labels it "sensitive" or "critical," there is a shared reference point that allows regulators to recognise they are governing the same underlying concern — enabling cooperation without requiring either side to abandon its domestic framework. The UNGD can focus on building interoperability between existing frameworks, enabling mutual recognition rather than demanding homogenization that disadvantages developing countries. (2) Leveling the field for the Global Majority: Due to concentrated supply chains and resulting knowledge asymmetry, most Global Majority countries possess limited capacity to build out AI governance mechanisms, independently assess AI risks, evaluate vendor claims, or develop evidence-based regulation. The UNGD can build pathways for the UN AI Scientific Panel to frame best practices and ensure that the Panel's outputs are actively translated into policy-relevant guidance accessible to under-resourced governments. (3) Enabling meaningful participation: The UNGD can focus on developing guidance and mechanisms around how to build multistakeholder processes to enable meaningful participation in AI governance processes, with a particular focus on ensuring that insights from civil society and academic organizations have a meaningful impact on the agenda and outcomes, as elaborated in answers to questions 8, 16 and 17.

-Leverage existing forums and work: The UNGD should advance its goals in coherence with the outcomes of the WSIS, as envisaged within the GDC and the WSIS+20 Outcome Document. It should also further their implementation by making recommendations to and building on the work of relevant bodies such as the CSTD, OHCHR and UNESCO. WSIS has also enabled the development of a network of National and Regional IGFs (NRIs), which can provide avenues for inclusive dialogue and bridge perspectives and regional realities on AI governance. The UNGD should also incorporate and implement the principles created through the NetMundial and NetMundial+10. -Learn from global processes: Map connections and learn from prior global processes of relevance, such as the history of internet governance as well as the global governance architecture that has emerged to address climate change, and has encouraged multilevel action from governments and private companies at the local, country, regional, and international levels to enact innovative policies suited to their specific responsibilities. The Intergovernmental Panel on Climate Change (IPCC) in particular has been integral for informing decentralized action. -Articulate a bottom up agenda: Move beyond episodic consultation to creating sustained infrastructure for equitable knowledge exchange to facilitate a bottom-up articulation of AI opportunities, priorities, and risks. By connecting with initiatives like the Multistakeholder Approaches to Participation in AI Governance (MAP-AI), Global South Alliance, Global Digital Rights Coalition, Global Digital Justice Forum, and others that are surfacing independent perspectives from the CSO and academic communities, the UNGD can strengthen multistakeholder participation and advance legitimate, empirically documented priorities and actions on AI governance. Intersessional working groups using open, participatory processes can also facilitate this sort of legitimate, grounded agenda and outcome development.

-Governments should provide regulatory perspectives and solutions as well as public policy priorities, especially from the Global Majority on developmental concerns, digital divides, and capacity constraints. -Private sector should share technical expertise and best practices alongside risk assessment mechanisms, while committing to transparency and accountability. -Civil society and academia should surface a bottom-up, evidence-driven agenda, representing affected communities, interrogating power asymmetries, advocating for accountability, and keeping the process grounded in human rights frameworks. Key formats: -Multistakeholder Advisory Group/Steering Committee: To provide guidance and bring necessary expertise and resources to the table, while helping to ensure representation as a fundamental value across UNGD. Another, complimentary approach called for by the GDRC, would be to appoint a "liaison to support the Co-Chairs to facilitate meaningful multistakeholder engagement ahead of, during, and following the Global Dialogue." -Thematic Balanced Multistakeholder Working Groups: These can be tasked with working through 2026 and part of 2027 (between the first and second UNGD's) to produce analysis that informs the Co-Chairs' outputs and helps drive concrete proposals, deliverables, and outcomes. -Multistakeholder regional consultations: For instance, by working through the IGF, NRIs, and leveraging its intersessional processes (Dynamic Coalitions, Best Practice Forums, Policy Networks, and the youth IGF initiatives). Key practices: -Continuous, robust consultation: Regular, open multistakeholder consultations, with fair opportunities for participation to shape the multilateral process and outcome. -Shared spaces and "mingled tracks" that bring together governmental and non-governmental stakeholders and work towards integrated outcomes from government, industry, civil society, and academia. Effective representation can be operationalized by having designated quotas for civil society and academic organizations and allocation of dedicated time for real-time dialogue between stakeholders in the agenda. -Ensuring transparent modalities and processes, including a digital platform for rolling submissions, recordings of meetings conducted, and public summaries explaining how stakeholder input informed outcomes.

Currently, multiple forums for internet governance and AI governance are dominated by select voices - states that have significant power, market size, and capacity, influential technology companies, and select Global North CSOs and academics. Although CSOs from the Global Majority participate in discussions, mere participation does not necessarily result in inclusion in agenda-setting, decision-making and framing of outcomes. The UNGD should be deliberately inclusive and enable effective participation of underrepresented and disadvantaged communities, especially those in the Global Majority. In addition to approaches to multistakeholder participation in our response to question 15, the UNGD can consider: Formats: -Citizen Track: Meaningfully include and highlight the voices of local communities, including through deliberative and public participation processes, as well as regional dialogues, including through the IGF's NRIs, collective intelligence collection, citizen representatives, and expert committees. Regional demographically representative focus groups/citizens' assemblies during the pre-deliberation stage to counterbalance the concentration of elite voices and highlight Global Majority concerns. Practices: -Multiple Avenues for Input: To enable effective participation of Global Majority communities, create a variety of avenues for participation and input, including written submissions, hybrid meetings, and regional and global consultations particularly at existing convenings in the Global Majority. -Intentionally Balanced Dialogue: Ensure balance between different stakeholder groups, such as the technical and scientific communities and socio-legal experts, and guard against privileging any one by enabling equal space across mechanisms to ensure democratic governance. -Travel Support: To be inclusive of diverse perspectives and acknowledging the reality of financial constraints for CSOs from the Global Majority, ensure support for in person participation by providing or seeking travel funds and facilitating visa support. -Language Accessibility: Given 3 billion people do not speak any of the UN official languages, to enable inclusion of communities, translate publicly available documents beyond the six UN languages.

The UNGD should adopt participatory and deliberative formats and practices that surface diverse perspectives and enable collaborative decision-making and draw on lessons from the rich history of multistakeholder Internet governance, as elaborated in our response to questions 8, 14, 15, 16 and 18. Importantly, the UNGD should incorporate and further the principles created through the NetMundial and NetMundial+10 enabling multistakeholder input and contribution to decision-making in agenda and outcome developments. This includes: -Multistakeholder Advisory Group: To provide guidance and bring necessary expertise to steer the agenda and outcomes. -Thematic, Balanced Working Groups: The work of the Co-Chairs can be supported by multistakeholder working groups on key themes tasked with working through 2026 and part of 2027 to produce analytical documents and insights that inform the Co-Chairs' outputs. -Focus Groups: Regional demographically representative focus groups or citizens' assemblies during the pre-deliberation stage. -Citizen track: Creation of Citizen track to enable bottom-up perspectives. Shared spaces and "mingled tracks" that bring together governments, companies, and civil society in deliberative spaces, enabling genuine, integrated input from government, industry, civil society, and academia. -Ensuring transparent modalities and processes, including a process roadmap, a calendar of consultations, a public repository of written submissions from various stakeholders, summaries explaining how stakeholder input informed final decisions and outcomes, and other opportunities for participation. -Robust Consultation: Formal outcomes from the UNGD should be shaped in multistakeholder forums before being input into the final multilateral adoption process. For instance, a published zero draft that allows stakeholders to respond to specific provisions creates accountability and enables the final document to reflect a wider range of inputs.

-Lessons from Internet Governance: For the UNGD to be truly inclusive, transparent, and meaningful, it should draw on lessons from the rich history of multistakeholder Internet governance- the WSIS and its Informal Multistakeholder Sounding Board (IMSB), as well as the Digital Rights Coalition's Five-Point Plan for an Inclusive WSIS+20 Review and Eight Practical Follow-up Recommendations; IGF, including the Multistakeholder Advisory Group (MAG) and Leadership Panel, and IGF processes identified under response to question 15; and the NETmundial process and São Paulo Multistakeholder Guidelines. -The UNGD can draw on lessons from NetMundial+10 that recognizes the importance of documenting how contributions were made, evaluated, and incorporated into the governance process, by publishing documentation explaining how written/verbal input was considered or rejected. Such transparency should not only be around input into outcome documents, but also on who contributes, is given an opportunity to participate and time allocated. -Mechanisms to address technological essentialism: Besides this, the challenges of disproportionate corporate and government voices and reliance on technological essentialism can be addressed through various mechanisms discussed in our responses like the creation of thematic working groups with representation from various stakeholder groups and a balanced mix of socio-legal and technical expertise, and a dedicated citizen track to highlight bottom-up perspectives and share best practices on the various themes of the UNGD and Focus Groups. Relevant issues should be shaped into outcomes. -Accountability structures: The UNGD should create an institutional architecture that enables accountability for the adoption and implementation of commitments among stakeholder groups, including government and industry to enable structural redistribution of decision-making power to communities most affected by AI systems. -Community feedback: Post-event, the UNGD should also facilitate community-level consultations which could be informed by surveys on process and structure, providing a feedback loop for future iterations of the UNGD.