Information Technology Industry Council (ITI)

We believe that a successful Global Dialogue will position the UN to play a constructive role in the global AI governance ecosystem by identifying gaps, convening stakeholders, and supporting coordination across existing efforts. Success depends on the establishment of a credible and inclusive multistakeholder forum where all stakeholders, including industry, can participate equally in discussions. This dialogue should ensure that AI governance discussions are grounded in evidence and real-world experiences. Governance frameworks that are disconnected from how AI is developed and deployed risk stifling innovation without meaningfully addressing harms. Points of emphasis include: • Commit to multistakeholder engagement. We urge the UN to keep the Dialogue genuinely multistakeholder, engaging the private sector, civil society, academia, and technical communities, not just as observers but as collaborative partners. • Prioritize key workstreams. While we appreciate the four thematic areas identified, numerous activities can be covered under each. If the UN plans to pursue specific work products or outcomes, we encourage it to use the first Dialogue meeting to identify a limited number of priority workstreams to retain focus and impact. • Clarify alignment with existing initiatives. As highlighted in the note, there are many ongoing initiatives outside of the UN focused on addressing different thematic areas in AI. The Dialogue should clearly articulate how it is enhancing these efforts, while keeping interoperability in mind. • Agree to undertake system-wide mapping. This will ensure that the UN is filling gaps with new or proposed work programs rather than duplicating efforts. The dialogue should assess AI adoption challenges, implementation realities, and best practices from diverse contexts. A successful dialogue would stress challenges presented by prescriptive frameworks that hamper organizations' innovative capacity. The summary should include an overview of the state of AI adoption internationally and best practices for the enablement of AI systems.

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

Safe, secure, and trustworthy AI is required to build trust and proliferate AI systems. Strong security foundations enable countries to adopt and scale AI with confidence. The UN is well positioned to drive consensus on the importance of security across the AI lifecycle. Emphasizing this will also help ensure emerging economies have the access and agency needed to develop, deploy, and use AI securely at scale. Where appropriate, developers should document and share high-level information about data curation and risk mitigation processes to build trust, provided such measures do not compromise trade secrets or security of AI systems. AI capacity building is essential to ensure that AI benefits reach all countries. Given its development mandate and universal membership, we believe the UN is especially well-suited to identify practical obstacles to AI adoption in emerging economies and foster information sharing across countries. Focusing on bridging AI divides complements the work being undertaken in other organizations. Interoperability and compatibility of AI governance approaches is another key priority as diverging AI frameworks become barriers to AI development and deployment, particularly for SMEs and innovators in developing countries. Different definitions, varying transparency requirements, and lack of conformity assessments create costs that often fall disproportionately on smaller businesses without the resources to comply. International standards and mutual recognition frameworks can build bridges across regulatory regimes, enabling compliance and supporting innovation. Governance frameworks should be risk-based, rather than imposing blanket requirements on all AI systems. Transparency, accountability and robust human oversight are needed to build trust necessary for broad AI adoption. Without clear accountability and human control mechanisms across the AI lifecycle, organizations cannot confidently deploy or rely on AI systems. Balanced transparency requirements and provenance mechanisms based on industry-led standards can help build trust across the ecosystem without imposing prescriptive or hard-to-scale obligations on developers.

There is a lack of coordination, continuity, and visibility across existing AI governance initiatives. While there has been significant progress made in multiple fora to address a wide range of AI-related issues, including through voluntary commitments, adoption of principles, and various outcome documents, these efforts often operate in parallel, with limited follow-up or opportunity for alignment. We believe the UN and the Global Dialogue can play a unique role in providing connectivity across these efforts, serving as a convener promoting interoperability amongst various domestic and regional approaches to AI governance, rather than creating redundant or competing mandates. A fragmented global landscape of overlapping commitments and obligations would burden innovators and slow AI adoption, particularly for smaller developers and developing countries. Another global issue is AI adoption. Adopting AI models is key for countries to realize the opportunities from AI across their economies. Governance outcomes from the Dialogue should consider supporting the AI ecosystem as a whole and educating governments and individuals on how advantageous AI adoption can be, while avoiding regulatory approaches that would disadvantage adoption and innovation. This dialogue should examine factors enabling or impeding responsible AI adoption, including infrastructure access, affordability, regulatory clarity and predictability, and the importance of interoperability and required skills and capacity development. Beyond this, open innovation and cross-border data flows are the bedrock to the global development of and access to AI systems, and this is a cross-sectoral priority. Restrictions that fragment the global data ecosystem undermine innovation and access, especially for developing nations.

One of the challenges of the current state of the global AI governance landscape is the proliferation of diverse initiatives across fora, with limited connection between them all. From an industry perspective, it can be difficult to know both how and where to engage, and it remains unclear which efforts will drive the most meaningful, practical impacts. While we appreciate the thematic tracks laid out, which mirror many tracks laid out in other fora, it is critical that UN work avoids overlapping with existing technical standards development processes, existing global AI principles, and regional legal frameworks, and instead prioritize areas where the UN's unique convening power and universal membership can have the most impact with a focus on enhancing connectivity across efforts and AI adoption as a key enabler. Relatedly, regulatory fragmentation creates specific compliance challenges for ITI members and their customers. We see significant opportunity for the UN to support institutional capacity-building and help to facilitate access to the resources needed to deploy AI responsibly and securely. Many countries face challenges in developing and implementing AI governance frameworks, participating in standards development processes, or accessing necessary resources to promote AI skilling, access, and adoption. As well, growing AI capacity gaps threaten to leave developing countries behind. This dialogue must prioritize concrete actions to bridge this divide, such as supporting digital infrastructure investment, free flow of data, the promotion of appropriate access to AI tools and training data and ensuring that governance frameworks do not hamper adoption for countries maturing their AI capabilities. These systems must be treated as a tool for advancing development, not as a technology to be restricted before its benefits are realized.

One of the challenges is the proliferation of diverse initiatives across fora, with limited connection between them all. From an industry perspective, it can be difficult to know both how and where to engage, and it remains unclear which efforts will drive the most meaningful, practical impacts. While we appreciate the thematic tracks laid out, which mirror many tracks laid out in other fora, it is critical that UN work avoids overlapping with existing technical standards development processes, existing global AI principles, and regional legal frameworks, and instead prioritize areas where the UN's unique convening power and universal membership can have the most impact with a focus on enhancing connectivity across efforts and AI adoption as a key enabler. Relatedly, regulatory fragmentation creates specific compliance challenges for ITI members and their customers. We see significant opportunity for the UN to support institutional capacity-building and help to facilitate access to the resources needed to deploy AI responsibly and securely. Many countries face challenges in developing and implementing AI governance frameworks, participating in standards development processes, or accessing necessary resources to promote AI skilling, access, and adoption. As well, growing AI capacity gaps threaten to leave developing countries behind. This dialogue must prioritize concrete actions to bridge this divide, such as supporting digital infrastructure investment, free flow of data, the promotion of appropriate access to AI tools and training data and ensuring that governance frameworks do not hamper adoption for countries maturing their AI capabilities. These systems must be treated as a tool for advancing development, not as a technology to be restricted before its benefits are realized. We believe these gaps are often compounded by the nature of the current governance landscape, where there are multiple multilateral initiatives underway and individual countries are also taking different approaches.

There are many ongoing AI initiatives that could be helpful for the UN to reference and/or seek to connect. We encourage the UN to consider how to best complement the ongoing global AI Summit series, including the upcoming Summit taking place in Switzerland in 2027. Each Summit has thus far had a slightly different focus; initially, the series was focused on safety, though more recently the Summit series has been broadened out to include AI adoption and democratizing access to AI for emerging economies. We believe the Global Dialogue could be especially useful in building upon the theme of democratizing access, which is critical for countries around the world to leverage the technology to support economic growth and make improvements in critical sectors/areas. We also encourage the Dialogue to explore and connect existing UN efforts, especially those taking place in UNESCO and the ITU. For example, considering how to leverage the ITU's AI for Good platform and how to best incorporate UNESCO's work on AI ethics would be useful. Technical standards efforts should be connected, particularly the ISO/IEC JTC 1/SC 42 on AI standards. These technical standards bodies develop frameworks through inclusive, consensus-based processes that can provide common frameworks for AI safety, security, transparency, and accountability. Such standards enable mutual recognition and regulatory interoperability across jurisdictions. Finally, the UN should build upon efforts being undertaken in the OECD's Global Partnership on AI. There are several efforts, in particular, that may be worth explicitly considering, including the OECD's AI Principles and the G7 reporting framework for the Hiroshima AI Process (HAIP) International Code of Conduct.

The structure of the Dialogue should prioritize coordination, inclusivity, and practical impact. Working groups or task forces should be carefully scoped to focus on cross-initiative coordination, rather than replicating the substantive work that is being undertaken in other fora. Similarly, any mechanisms that are developed related to pledges or commitments should emphasize transparency and provide a meaningful way to engage in follow-up discussion, rather than creating parallel or additional processes. To be sure, there have been several batches of commitments that have come out of both the G7, G20, and the global AI summits that have not been revisited or progressed. We appreciate that the UN is consulting widely on the shape and structure of the Dialogue via webform, as this is one way to include a variety of voices who may otherwise not be able to participate. The UN should seek to be as inclusive as possible, ensuring that Member States, civil society, academia, and industry are represented. In considering how to best include business, we encourage the UN to ensure that a variety of technology industry stakeholders are represented from across the AI tech stack. Trade associations can help bring the perspectives of many different stakeholders together and articulate a consensus view of the industry. It may also be useful to consider including sector-specific stakeholders if there are specific conversations that would benefit from better understanding how AI is being deployed and/or used in different verticals. The AI Dialogue should prioritize interactive, engagement over formal presentations, favoring collaborative discussions and problem-solving, real-world case study deep-dives, and technical demonstrations of AI systems.

As mentioned above, the best outcomes result from processes that include diverse global stakeholders. In multilateral initiatives, small companies, start-ups, civil society, and local populations have not meaningfully been involved. The UN should be intentional and proactive with reaching diverse communities that may not have previously engaged with the UN. The UN should refer to recommendations developed at NETmundial+10 in 2024 for how to effectively engage across sectors for digital policy processes. SMEs are underrepresented in AI governance discussions despite being disproportionately affected by regulatory fragmentation. Broader industry participation with cloud infrastructure providers, semiconductor manufacturers, AI labs, software developers, systems integrators, and sectoral AI users is required. Developing country innovators and AI practitioners need stronger representation, as governance frameworks developed primarily by and for advanced economies may not address challenges specific to developing country contexts, such as limited infrastructure, different risk profiles, or distinct social and economic priorities. Researchers from diverse disciplines beyond computer science bring important perspectives on AI's social, economic, and cultural implications. Civil society organizations representing affected communities, workers, consumers, and marginalized groups provide essential perspectives on AI risks and impacts that elude developers and policymakers.

We believe the initial format proposed for the first in-person meeting makes sense as it offers a mix of different types of sessions. However, to also move activity forward in the downtime between official Global Dialogue meetings, the UN should consider whether multi-stakeholder roundtables tied to the specific thematic areas/priorities, regional consultations, or capacity-building workshops would be helpful. We also believe that providing mechanisms for ongoing written input could help ensure continuity between Dialogue meetings. Potential roundtable topics include real-world use case deep-dives that select key AI deployments and examine them holistically to base governance discussions on reality. Also, standards development workshops that provide engagement with how technical standards are created, assisting policymakers' understanding of how standards support regulatory objectives, what standards currently exist, and what is on the horizon.

There are many different types of policies, practices, and approaches that can inform effective AI governance. As we've mentioned, we encourage the UN to reference and build upon these approaches. For example, we suggest that the UN look to the OECD AI Principles and determine how to build upon these, incorporating a wider set of perspectives from Member States. The OECD's AI Observatory could also be a helpful tool in informing the UN's understanding of the global AI governance landscape. NIST's AI Risk Management Framework offers a helpful way for organizations to begin operationalizing risk management, which is a critical part of AI governance. There are international standards that can also help to support risk management and governance, such as ISO/IEC 42001. We also highlight ITI's AI Accountability Framework, which outlines practices that ITI and our members believe are critical to operationalizing the responsible development and deployment of AI. Finally, we encourage the UN to build public-private partnership mechanisms, incorporating industry and civil society to enable and increase AI skilling, access, and adoption by new populations and communities.