United Nations Dept of Global Communications

A successful first Global Dialogue on AI Governance would: 1. Recognize information integrity as a foundational, cross-cutting issue for AI governance. AI systems now mediate how information is created, surfaced, ranked, recommended, translated, and monetized across the platforms and services on which billions of people depend, with systemic implications for the right to seek, receive and impart information, for the safety of children and other at-risk groups, and for international cooperation on shared global challenges. 2. Anchor discussions in multilateral instruments and frameworks for action, including the Global Digital Compact, the Pact for the Future, international human rights law and the UN Global Principles for Information Integrity (GPII), so that interoperability is built on universally agreed terminology rather than concepts imported from particular regional regulatory regimes. 3. Address AI's effects on the information ecosystem at the systemic level, including business model incentives, platform design, advertising supply chains, recommender and generative systems, and the sustainability of independent media, alongside individual model-level concerns. 4. Open clear pathways for inclusive participation – particularly for civil society, independent media, fact-checkers, researchers, children and youth, affected communities and information defenders from across geographies. 5. Identify concrete deliverables for the 2027 session in New York, including a multi-stakeholder workstream on interoperable standards for AI-generated content provenance and disclosure, and a capacity-building track that recognises information integrity capacity as a core dimension of bridging AI divides. Above all, success means ensuring that, as AI becomes embedded in the information ecosystem, its governance serves the public interest, upholds human rights, is solidly grounded in principles of transparency and accountability and supports international cooperation on shared global challenges.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

This selection reflects the cross-cutting nature of information integrity in AI governance. Information integrity refers to an information ecosystem in which reliable and accurate information is available to all, enabling people to engage meaningfully in public life, make informed decisions, and exercise their rights. Strengthening information integrity means protecting the right to freedom of expression and access to information, ensuring inclusive access to a range of information sources and enabling people to navigate information spaces safely, with privacy and freedom. It involves building resilient societies that foster trust, knowledge and public empowerment. Challenges to information integrity encompasses a spectrum of risks, such as disinformation, hate speech, restrictions on press freedom and the malicious use of technologies. AI systems are now central to that ecosystem, and the four areas selected are where their effects are most acute. Protection and promotion of human rights is foundational. AI systems that mediate information engage Article 19 obligations on freedom of expression and the right to seek, receive and impart information. Established frameworks - including the Rabat Plan of Action - provide a principled basis for assessing the proportionality of restrictions on expression, including those mediated by AI. Protection and promotion of human rights is foundational, in particular the right to freedom of expression and to seek, receive and impart information. Established frameworks provide a principled basis for assessing these issues. The UN GPII anchors information integrity in international human rights law, including freedom of expression and access to information standards. Transparency, accountability, and human oversight operationalize these rights. The GPII identifies transparency and research as a core pillar, calling for meaningful researcher access, accountability mechanisms accessible to affected populations, and effective remedies. These apply directly to recommender systems, generative AI in search and information services, and AI-driven advertising. Social, economic, ethical, cultural, linguistic and technical implications captures the systemic dimensions of information integrity: AI's effects on the digital advertising business models that finance much of the public information environment; on the sustainability of independent and local media; on linguistic equity for speakers of low-resource and Indigenous languages; and on populations facing disproportionate risks, including children, women in public life, journalists, scientists, and human rights defenders. Safe, secure and trustworthy AI must encompass risks to information environments - including coordinated manipulation, generative content at scale, and the erosion of source provenance - alongside individual model-level concerns. Without this framing, "safety" risks being defined too narrowly to capture the systemic harms that most affect the public.

Several cross-cutting and emerging issues warrant explicit recognition by the AI Dialogue: Information integrity is the most significant cross-cutting issue. The integrity of the global information ecosystem is under increasing strain, and the rapid deployment of AI is intensifying that strain. The Global Digital Compact, the UN Global Principles for Information Integrity (2024) and related initiatives provide the framework for addressing these risks. The UN GPII is organised around five pillars: societal trust and resilience; healthy incentives; public empowerment; independent, free and pluralistic media; and transparency and research. Each pillar is engaged by AI development and deployment, and each is relevant to every thematic cluster identified by the General Assembly. The economics of AI and digital advertising. AI is being rapidly integrated into the digital advertising systems that finance much of the public information environment. This affects the sustainability of independent media, the incentives that shape platform design, and the conditions under which AI itself is being commercialised. The GPII pillar on healthy incentives speaks directly to this issue. Children's rights in AI-mediated environments. AI-generated child sexual abuse material, manipulative design targeting minors, and AI companions raising age-appropriate concerns require dedicated multilateral attention, in line with General Comment No. 25 of the Committee on the Rights of the Child. Safety of journalists, researchers, scientists, and human rights defenders in AI-mediated environments - including AI-enabled harassment, image-based abuse, and gendered disinformation - with serious chilling effects on freedom of expression. AI's role across high-stakes thematic contexts - including elections, public health, climate, and conflict - where AI-driven manipulation of information environments has direct consequences for human rights, peace

As the United Nations entity coordinating the UN system's work on information integrity, DGC observes systemic challenges and emerging opportunities across all regions: Challenges. Generative AI is lowering the cost of producing convincing disinformation at scale, in growing volumes and increasing sophistication, including in synthetic media. AI-driven recommender systems and advertising tools shape the information environments in which billions of people encounter information about elections, public health, climate, and conflict, with limited transparency. Researchers, journalists, scientists, and human rights defenders working on these issues face AI-enabled harassment and gendered abuse, with documented chilling effects. Linguistic and contextual asymmetries leave many countries and contexts with degraded access to safe, accurate AI-mediated information in their own languages and cultural contexts. Capacity to assess AI's effects on domestic information environments remains uneven, and researcher access to data and models is structurally constrained. These challenges are systemic. They cannot be resolved by content-level interventions alone, nor by national action in isolation. They require the international cooperation the AI Dialogue is designed to advance. Opportunities. The UN Global Principles for Information Integrity, launched by the Secretary-General in June 2024, provide a shared normative framework that has gained significant traction. The Global Digital Compact and the Pact for the Future have anchored commitments to information integrity in agreed multilateral language. DGC's operational methodology — the 3R framework (Research, Risk Assessment, Response) — translates the GPII into practice, organiszd across four response objectives: prevention, protection, mitigation, and recovery. Multi-stakeholder coalitions in priority thematic areas (including climate, elections and public health) demonstrate the value of coordinated action across Member States, the UN system, civil society, and the private sector. These models offer transferable lessons for the AI Dialogue's work.

The AI Dialogue can play a unique and necessary role in three areas: First, normative coherence. A patchwork of incompatible regional regulatory approaches is emerging. The Dialogue is uniquely placed to anchor international cooperation in agreed multilateral instruments - including the Global Digital Compact, the Pact for the Future, WSIS and international human rights law — rather than in terminology drawn from particular regional regimes. This is essential for genuine interoperability and for ensuring that AI governance serves countries at all stages of digital development. Second, integration with existing global priorities. AI does not exist in isolation from the major challenges the international community is working to address — the protection and promotion of human rights, the safety of children, the conduct of elections, public health, climate action, the sustainability of independent media, and the realisation of the Sustainable Development Goals. The Dialogue can draw connections between AI governance and these priorities, ensuring that AI-related discussions reinforce — rather than fragment — the broader multilateral agenda. Third, inclusion. Many AI governance discussions to date have been concentrated in a small number of countries and sectors. The Dialogue can expand the conversation through structured participation for civil society, independent media, fact-checkers, researchers, and Global Majority voices. The UN's experience with multi-stakeholder coalitions on information integrity shows that broad participation, anchored by Member States and supported by international organisations, can shift the centre of gravity on global issues and translate principles into practice. Beyond these structural roles, the Dialogue can also serve as a vehicle for sharing concrete good practices and translating high-level principles into operational tools that Member States can adapt to their own contexts.

The AI Dialogue should connect with, and add value to, a broad set of existing initiatives: UN frameworks and processes. The Global Digital Compact and the Pact for the Future; the UN Global Principles for Information Integrity (2024);; the WSIS+20 review; OHCHR's work on AI and human rights, including the application of established frameworks such as the Rabat Plan of Action; UNESCO's work on AI ethics, media viability, and platform governance, including the Guidelines for the Governance of Digital Platforms; UNICEF's work on AI and children's rights, including General Comment No. 25; and ITU's standardization work. DGC-led mechanisms. The UN System-Wide Working Group on Information Integrity, chaired by DGC, coordinates approaches across the UN system. A planned Global Network for Information Integrity will connect researchers, practitioners, civil society, and UN personnel to match expertise with country-level needs. Multi-stakeholder coalitions. UN-convened coalitions in priority thematic areas — including the Global Initiative for Information Integrity on Climate Change, co-led by the UN, Brazil and UNESCO — demonstrate how information integrity and the GPII can be operationalized across thematic domains. The Charter for Accountable Climate Advertising, being developed under the GIIICC, demonstrates how multilateral commitments can be operationalized with the private sector. Standards bodies and provenance initiatives. The Dialogue can support coherence between technical work relevant to AI-generated content and human-rights-based governance. Added value. The Dialogue can connect these efforts, ensure AI governance is rooted in multilateral frameworks, and translate principles into concrete, interoperable practices.

The format and structure of the AI Dialogue should reflect its role as an inclusive, multi-stakeholder platform: Multi-stakeholder coalitions and working arrangements can play a substantive role, not only an advisory one. The UN's experience with information integrity coalitions across thematic areas — anchored by Member States, co-led with UN entities, and supported by Advisory Groups drawn from civil society, academia, media, and the private sector — shows how broad participation can be translated into concrete deliverables. Comparable arrangements could give civil society, researchers, and other stakeholders meaningful pathways to contribute to the AI Dialogue's substantive work. Civil society, independent media, fact-checking organisations, and researchers should have dedicated speaking time and structured pathways into thematic preparations, not solely observer status. Their evidence and lived experience are essential to credible AI governance. The private sector — including AI developers, platforms, advertisers, and the digital advertising supply chain — should engage in line with the UN Guiding Principles on Business and Human Rights, with particular attention to the role of advertising business models in shaping the information environment. Stakeholders from developing countries should be supported to participate fully, including through capacity-building support and accessible documentation in multiple languages. Practical recommendations for format: Thematic preparatory consultations ahead of each session, with structured stakeholder input. Hybrid participation to enable engagement from across regions and time zones. Dedicated thematic clusters with named co-leads from Member States and stakeholders. A standing pathway for affected populations — including children, women in public life, journalists, scientists, and human rights defenders — to share lived experience. Co-Chairs' summaries that draw explicit connections to existing UN frameworks and initiatives, ensuring continuity between the 2026 and 2027 sessions.

Several communities and perspectives remain underrepresented in global discussions on AI governance: Geographic limitations. AI governance discussions to date have been concentrated in a small number of countries and institutional settings with stakeholders from many developing countries systematically underrepresented — despite often facing the most acute challenges from AI-mediated information environments, including in low-resource and Indigenous languages. Independent media, journalists, and fact-checkers. Those whose work is most directly affected by generative AI and AI-driven distribution systems are rarely at the table when their conditions of work are discussed. Children and young people. Despite being among the most affected by AI-mediated environments, their perspectives are rarely structured into international AI governance, contrary to the participation rights affirmed in General Comment No. 25 of the Committee on the Rights of the Child. Women in public life, particularly those facing AI-enabled gendered abuse and image-based abuse, including deepfake harassment. Researchers, scientists, and human rights defenders working on contested topics, who face AI-enabled harassment and disinformation campaigns with documented chilling effects on freedom of expression and the conduct of independent inquiry. Indigenous Peoples, whose languages, cultural knowledge and rights are directly affected by AI training data and outputs. Inclusion mechanisms. The AI Dialogue should: provide travel and translation support for Global Majority participation; convene structured consultations with affected communities ahead of each session; partner with networks that already represent these communities (including UN-convened coalitions); and ensure that Co-Chairs' summaries reflect underrepresented perspectives, not only the views of well-resourced participants.

Several engagement formats can foster meaningful and dynamic participation: Thematic working sessions co-led by Member States and stakeholders. Rather than panel-and-audience formats, the Dialogue can convene smaller thematic sessions — for example, on AI and information integrity, on AI and children's rights, on AI and elections, or on AI in advertising and media — with structured contributions from governments, UN entities, researchers, civil society, and affected communities. The UN has used this approach successfully to advance work on information integrity in priority thematic domains. Lived-experience segments. Dedicated time for journalists, scientists, fact-checkers, women facing AI-enabled abuse, and young people to share experience of AI-mediated harms — informing, not interrupting, the substantive discussion. Demonstration showcases. Short, structured presentations of concrete good practices — from country-level implementation of GPII-aligned approaches, to UN-convened multi-stakeholder coalitions, to provenance and disclosure standards being developed by industry and standards bodies. These showcases can move the Dialogue from principles toward practical models. Pre-session multi-stakeholder consultations, regionally anchored, with outputs feeding directly into thematic discussions in Geneva and New York. Hybrid participation should be available throughout. A standing youth participation track, in line with General Comment No. 25, with structured pathways for young people to contribute to thematic preparations and to plenary discussions. Cross-thematic dialogues. Information integrity, child safety, elections, public health, and human rights are interconnected. Sessions that explicitly bridge these areas — bringing together negotiators, researchers, and practitioners from different communities — can surface insights that single-track discussions miss.

Examples of approaches that promote effective governance of AI in the global information ecosystem: The UN Global Principles for Information Integrity (GPII), launched by the Secretary-General in June 2024, set out a normative framework organised across five pillars: societal trust and resilience; healthy incentives; public empowerment; independent, free and pluralistic media; and transparency and research. The GPII recognizes advertisers and AI developers as key stakeholders with the power to strengthen - or undermine - the information ecosystem, and offer a shared reference point for AI governance discussions. A new issue brief on Advertising, AI and the global information crisis (April 2026, co-authored with the Conscious Advertising Network) examines how AI is being integrated into the digital advertising systems that finance much of the information environment, and identifies risks and recommendations against each of the five GPII pillars. It is a model for thematic, evidence-based UN policy guidance on AI and information integrity. The 3R operational methodology - Research, Risk Assessment, Response - set out in DGC's issue brief From Principles to Practice: Strengthening Information Integrity, translates the GPII into practice. Response measures are organised across four objectives: prevention, protection, mitigation, and recovery. Multi-stakeholder coalitions. The Global Initiative for Information Integrity on Climate Change, co-led by the Government of Brazil, the UN, and UNESCO, demonstrates how the GPII can be operationalised in a specific thematic domain through coalition-building, normative anchoring, and concrete commitments. The Belém Declaration on Information Integrity on Climate Change, launched at COP30 in November 2025 and signed by 26 countries and the EU, offers a concrete model of multilateral cooperation on information integrity in the AI era. The Charter for Accountable Climate Advertising, under development through the GIIICC, illustrates how multilateral commitments can be operationalized with the private sector.