What is the UN Global Dialogue on AI Governance?
AI regulation: concepts, institutions and standards
As of 4 June 2026, the UN Global Dialogue on AI Governance is a non-binding UN forum, mandated by the Global Digital Compact and established by General Assembly resolution A/RES/79/325, where governments and other stakeholders discuss international co-operation on AI. It is not a treaty or global regulator. Its first full session is scheduled for Geneva on 6 to 7 July 2026, and it is designed to inform future common understandings on priority AI governance issues.
What this means
The Dialogue is the UN's global discussion forum for AI governance in the non-military domain. It is the place where all UN Member States, plus relevant stakeholders, can compare approaches, flag risks, share practice and discuss how international co-operation on AI should work.
It is not the same as the Global Digital Compact. The Compact is the 2024 political agreement that called for the Dialogue. The Dialogue is the forum built to carry that commitment into practice.
It is also not the same as the Independent International Scientific Panel on AI. The Panel prepares independent assessments. The Dialogue is the political and multi-stakeholder space that receives that evidence, debates priorities and produces co-chair summaries that can shape later UN discussion.
Why it matters
For organisations that build, buy, deploy or govern AI across borders, the Dialogue matters because it can influence the language of international expectation before that language hardens elsewhere. The themes now shaping the Geneva session, safety, interoperability, human rights, transparency, accountability, human oversight, access, capacity-building and open models, overlap with issues already surfacing in procurement, board oversight, assurance and cross-border compliance.
It also matters geopolitically. Unlike many AI forums, the Dialogue gives smaller and lower-capacity states a formal place to press for access, skills, infrastructure and development priorities, not only risk controls drafted by larger AI-producing powers. If those priorities gain traction, buyers, vendors, advisers and public agencies may need to show not just technical safety, but also how their governance approach travels across jurisdictions and different levels of state capacity.
In practical terms, leaders should treat the Dialogue as a horizon-scanning and alignment forum. It does not replace domestic law, but it can affect how governments, UN agencies and adjacent international bodies talk about AI co-operation, especially ahead of the 2027 New York session and the later high-level review of the Global Digital Compact.
How it works
Mandate and legal status
The Dialogue sits in the UN's soft-law and co-ordination layer. Heads of State and Government agreed in the Global Digital Compact, adopted in 2024, that the UN should initiate a global dialogue on AI governance. The General Assembly then gave that commitment concrete institutional form through resolution A/RES/79/325, adopted by consensus on 26 August 2025.
That makes the mechanism real and formally anchored inside the UN system. It does not, however, make it binding in the way a treaty, regulation or convention would be. The Dialogue does not create direct duties for companies, does not impose sanctions and does not override domestic law or regional instruments. It is a forum for discussion, co-ordination and agenda-setting. The resolution also keeps the mechanism in the non-military domain, so it is not the UN's main track for military AI.
Who runs it and who can take part
The participation model is universal by UN standards. All Member States are invited to participate, and the process is also open to relevant stakeholders, including the private sector, academia, the technical community and civil society.
For the 2026 cycle, the President of the General Assembly appointed the Permanent Representatives of El Salvador and Estonia as co-chairs. UNESCO has described the first Dialogue's joint secretariat as comprising the International Telecommunication Union, UNESCO, the UN Office for Digital and Emerging Technologies and the Executive Office of the Secretary-General. That is important institutionally: the Dialogue is designed to draw on existing UN capacity rather than become a freestanding regulator.
What the Dialogue covers
The General Assembly identified seven substantive areas for the Dialogue. In plain terms, they cover safe, secure and trustworthy AI systems; capacity gaps and capacity-building; the wider social, economic, ethical, cultural, linguistic and technical implications of AI; interoperability and compatibility across governance approaches; respect for human rights; transparency, accountability and robust human oversight; and the development of open-source software, open data and open AI models.
For the 2026 Geneva session, the co-chairs grouped those issues into four working clusters: AI opportunities and implications; bridging AI divides; safe, secure and trustworthy AI with interoperable governance approaches; and human rights, transparency, accountability and human oversight. The official note on themes and structure makes clear that this grouping is an organising device for the meeting, not a rewrite of the mandate already agreed by Member States.
How the process is organised in practice
The Dialogue is intended to meet annually, usually for up to two days, in the margins of existing UN conferences and meetings. As of 4 June 2026, the first full session is scheduled for Geneva on 6 to 7 July 2026, and a second session is planned for New York in May 2027.
The 2026 Geneva programme is expected to include a high-level segment, thematic sessions and side events. The UN also ran a preparatory process around the meeting: consultations with Member States, stakeholder consultations, written submissions, an official call for side events and travel support for some eligible participants from developing countries. That operational detail matters because the Dialogue is not only a conference stage. It is a wider consultation cycle that shapes what reaches that stage.
What evidence and records it creates
The Dialogue is not itself a scientific assessment body. Instead, it is designed to receive evidence from the Independent International Scientific Panel on AI, which produces independent assessments of AI's opportunities, risks and impacts and presents its annual report at the Dialogue. In short, the Panel is the evidence-producing mechanism, and the Dialogue is the deliberative mechanism.
What the Dialogue produces is also narrower than many readers assume. Each session ends with a co-chair summary, not a negotiated global code or treaty text. The point of that design is to let all countries participate on equal footing without forcing every meeting into line-by-line diplomatic negotiation. For operators and governance teams, the practical record is the panel report, the agenda notes, written inputs, side-event material and the co-chair summary. Those materials are useful for horizon scanning and policy comparison, but they are not audit standards or compliance badges.
How it connects to future UN discussion
The adopted resolution gives the second Dialogue a more political follow-up role. Its co-chairs are expected to hold intergovernmental consultations to agree common understandings on priority areas for international AI governance, taking account of the earlier co-chair summaries and stakeholder contributions. In plain English, the Dialogue is meant to feed later UN debate through shared problem statements and priority areas, not by instantly generating binding rules.
That makes the 2026 Geneva and 2027 New York sessions more than standalone diplomatic events. Together, they form the main paper trail for what the General Assembly may do next when it reviews implementation of the Global Digital Compact. The Assembly also kept open the possibility of revisiting the Dialogue's terms of reference and continued life during the Compact's later high-level review, so the mechanism is established but not yet institutionally final in every respect.
Examples
Example one: a trade association, research group, non-profit or company with a concrete governance point could submit written input to the UN process before the portal closed, then use the official stakeholder consultation track to react to the proposed themes for Geneva. That is the clearest formal route for trying to shape the first session before the co-chairs write their summary.
Example two: an organisation seeking a visible role in Geneva could apply to run a side event under the UN's open call. The joint secretariat reviews proposals in consultation with the co-chairs, and the first session is expressly designed to include side events alongside the formal programme. That does not change the legal mandate, but it does create a structured way to place practical co-operation ideas in front of delegates and other participants.
Example three: a delegate or stakeholder from a developing country who would struggle to reach Geneva could use the UN's participation-support route, which in 2026 included an official travel-support application process. In July, that person can take part in the same general forum that will hear the Scientific Panel's preliminary report and feed the first co-chair summary into the wider path toward common understandings before the 2027 New York session.
Common misunderstandings
It is a UN treaty or a global AI law. No. It is a UN forum established by General Assembly resolution, with no direct binding force on states or companies.
It is the same thing as the Global Digital Compact. No. The Compact set the political commitment. The Dialogue is one of the mechanisms created to implement that commitment.
It is the same thing as the Independent International Scientific Panel on AI. No. The Panel produces independent assessments. The Dialogue is the forum that receives those assessments and discusses governance.
It replaces the OECD, UNESCO, G7, Council of Europe or regional AI frameworks. No. The UN presents it as a complementary universal forum, not a substitute for existing instruments or institutions.
It covers military AI. No. The mechanism is limited to the non-military domain, while military AI is being handled in other UN tracks.
Risks and boundaries
The biggest risk is over-reading the Dialogue. Attendance in Geneva, a side event slot or a favourable co-chair summary does not create legal compliance, safe deployment or market access. Domestic law, sector rules, procurement terms and existing international instruments still do that work.
A second boundary is the nature of the deliverable. A co-chair summary is not negotiated text. It can record themes, areas of alignment and live disagreement, but it does not bind states. Even the next planned step, intergovernmental consultations around the second Dialogue on common understandings in priority areas, stops well short of a treaty or automatic UN standard.
A third limit is timing. As of 4 June 2026, the first full working session has not yet happened. So the forum's practical weight, drafting style and influence still need to be tested in practice. The General Assembly also reserved the option to revisit the Dialogue's terms of reference and continued life during the later review of the Global Digital Compact at its eighty-second session.
Finally, scope matters. The mechanism does not address military AI, and it is not an assurance scheme, certification body or audit framework. Organisations should use it to track soft-law direction and international alignment, not as a substitute for legal analysis, technical testing or sector-specific governance work.
What to do next
Track the July 2026 Geneva session closely. In practice, the most useful items to watch are the Scientific Panel's preliminary report, the final programme, the themes that receive the most political attention and the co-chair summary that follows.
Map your organisation's current governance approach against the Dialogue's substantive areas: safety, interoperability, human rights, transparency, accountability, human oversight, capacity-building, wider societal effects and open models. That does not mean adopting a UN script. It means checking where your current work is likely to align cleanly with international discussion and where you may face pressure to explain gaps.
Decide whether direct engagement is worth the effort. If you operate globally, work with public-sector clients, depend on cross-border trust or sell governance-sensitive AI systems, it may be sensible to engage through a trade body, standards network, civil society coalition, research institution or side-event partner rather than waiting for second-hand summaries.
Keep expectations realistic. The Dialogue is a policy and co-ordination forum, not a shortcut around existing law. Use it alongside, not instead of, closer reading of binding regimes, sector supervision, procurement duties and technical assurance work. Then plan a second check-in before the May 2027 New York session, because that is the point at which the UN process is expected to move from broad exchange toward common understandings on priority areas.
FAQs
Is the UN Global Dialogue on AI Governance legally binding?
No. It is a UN forum created by General Assembly resolution, not a treaty or directly enforceable code.
Who can participate in the Dialogue?
All UN Member States are invited, and the process is also open to relevant stakeholders such as private-sector actors, academia, the technical community and civil society.
Is the Dialogue the same as the Global Digital Compact?
No. The Global Digital Compact is the 2024 political agreement that called for the Dialogue. The Dialogue is one mechanism created to implement that commitment.
Is the Dialogue the same as the Independent International Scientific Panel on AI?
No. The Panel prepares independent assessments. The Dialogue is the UN forum that receives that evidence and discusses governance priorities.
What does the Dialogue actually produce?
Each session produces a co-chair summary. The broader record also includes the Scientific Panel's report, written submissions, agenda notes and side-event material.
Does the Dialogue replace national or regional AI regulation?
No. It may influence international expectations and policy alignment, but domestic law and regional instruments still govern legal duties and enforcement.
Does the Dialogue cover military AI?
No. The mechanism is limited to the non-military domain. Military AI is being discussed in other UN channels.
What are the next major dates to watch?
The first full session is scheduled for Geneva on 6 to 7 July 2026. The second session is planned for New York in May 2027, ahead of the later General Assembly review of the Global Digital Compact.