Closing: the way forward — UN Global Dialogue on AI Governance

Session: Closing sessions of the inaugural Global Dialogue, including the high-level plenary, the “Dialogue of Dialogues” reflections, the four thematic cluster report-backs, and the closing ceremony
Scale: More than 4,200 registered participants from about 1,800 entities, representing nearly 170 member states
Outcome: A non-binding co-chairs’ summary (to be issued after the session), carried forward with the scientific panel’s report
Next: Second Global Dialogue, New York, May 2027

What the closing did

The closing sessions gathered two days of debate into conclusions and next steps rather than a negotiated text. No outcome document was adopted on the floor. Instead the two co-chairs, Ambassadors Egriselda López of El Salvador and Rein Tammsaar of Estonia, will prepare a co-chairs’ summary that, together with the scientific panel’s report, feeds the next Dialogue. The framing throughout was continuity: the inaugural gathering was treated as the opening of a recurring process, not a one-off event.

A co-chair, in closing: “This is not the end of a process. It is the beginning of one.”

The “Dialogue of Dialogues”

The high-level reflections, moderated by Claire Melamed (Vice-President for AI and Digital Cooperation, UN Foundation), brought together Halla Tómasdóttir (President of Iceland); Brad Smith (Vice Chair and President, Microsoft); Jan Tallinn (co-founder, Future of Life Institute); Gaia Marcus (Director, Ada Lovelace Institute); and Guy Ryder (UN Under-Secretary-General for Policy). The through-line was the gap between how fast AI is advancing and how slowly governance is keeping up, and the widening divide between the global North and South.

Halla Tómasdóttir, President of Iceland: “The central question is not simply how we govern artificial intelligence, it is whether we can govern ourselves in the age of artificial intelligence.”

Brad Smith put the pace problem in blunt terms, and Jan Tallinn pressed the point that the newest systems no longer behave like ordinary tools.

Brad Smith, Vice Chair and President, Microsoft: “Over the last 12 months, AI has raced forward, and AI governance has not kept pace.”

The four clusters report back

Each thematic cluster reported to the plenary in two minutes, distilling two days of discussion into a single take-away.

Cluster 1 (opportunities and implications). Reported by Mark-Alexandre Doumba (Minister of Digital Economy and Innovation, Gabon) and Rashid Khan (co-founder, Yellow.ai). The message: do not race to be first, race to use AI wisely; adapt “big AI” to “small AI” in local languages; capacity, not access, is the finish line.
Cluster 2 (bridging divides). Reported by Jovan Kurbalija (Executive Director, DiploFoundation) and Senegal. The divides are access, participation and skills; bridging them “goes beyond AI” and needs connectivity, electrification and investment first.
Cluster 3 (safe, secure and trustworthy AI). The co-chairs reported that governance must be adaptive, interoperability must become practical (shared definitions, standards, cross-border sandboxes, incident reporting), and accountability must stay clearly assigned so “autonomy never becomes an accountability gap.”
Cluster 4 (human rights). Linda Bonyo (founder and CEO, Lawyers Hub) reported for the cluster, pressing child protection, financing for human rights, the missing voices of AI workers, and the visa refusals that keep global-South participants away. The cluster proposed an “AI passport,” and Bonyo’s organisation launched an AI Governance Index for Africa.

Mark-Alexandre Doumba, Minister, Gabon: “Do not race to be first on AI, race to use AI wisely.”

What was agreed and carried forward

Because the Dialogue produced a summary rather than a declaration, the concrete outcomes are the commitments and mechanisms that now roll into the next cycle.

A date and a rhythm. The Second Global Dialogue will be held in New York in May 2027; Guy Ryder framed the Dialogue as a recurring, roughly annual place to convene.
The outcome is a co-chairs’ summary, not a negotiated declaration.
The scientific panel delivered. The Independent International Scientific Panel on AI delivered its first report, organised in seven working groups; a second report is expected before the New York session, and one cluster proposed aligning the panel’s working groups to the Dialogue’s four clusters.
The enhanced UN AI Resource Hub was launched (capacity-building and fellowships), now hosting 1,000+ AI initiatives from 60+ UN entities, mandated by paragraph 86 of the WSIS+20 outcome document.
A capacity-building network is going operational. A UN-supported global network of centres for AI capacity-building is entering its operational phase, with member states invited to nominate centres of excellence.
The Secretary-General’s Day-1 proposals were carried forward: a child-safety pledge, a transparency initiative, and a proposed Global Fund for AI (explicitly backed by several delegations).

The closing ceremony featured Guy Ryder, Tomas Lamanauskas (Deputy Secretary-General, ITU), Gabriela Ramos (Assistant Director-General, UNESCO), Amandeep Singh Gill (UN Under-Secretary-General and the Secretary-General’s Envoy on Technology), and the two co-chairs.

Jan Tallinn, Future of Life Institute: “These systems are actually no longer tools awaiting instruction, and our intuitions and institutions are not ready for machines that decide.”

Why it matters for the SDGs

The Dialogue is the implementation vehicle for the Global Digital Compact’s three lines of action (governance, scientific evidence, and capacity-building), all under the Pact for the Future. That maps to SDG 17 (means of implementation, technology, partnerships), while the North-South diffusion gap and the capacity-building network map to SDG 9 and SDG 10. The measurement angle SDGCounting watches: the recurring call for a shared, independent evidence base and comparable evaluation is what will let anyone judge whether AI governance is actually working.

Watch & read

UN Web TV, recording of the Day-2 plenary and closing sessions (7 July 2026).
Independent International Scientific Panel on AI.
Full Global Dialogue coverage.

Quotations are lightly edited from an automated (Otter.ai) transcript of the UN Web TV recording and should be read as close paraphrase; names and titles were reconciled to public records and reflect roles at the time.