FREE. FAIR. GLOBAL.
Planning the Second UN Dialogue on AI Governance
UN Member States began preparing for the May 2027 AI governance dialogue, raising questions about frontier AI risks, human oversight and how developing countries can participate.
Katherine Knisely

November 19, 2026 · United Nations Headquarters
UN Member States began preparing for the May 2027 AI governance dialogue, raising questions about frontier AI risks, human oversight and how developing countries can participate.
Artificial intelligence is advancing quickly, and delegations raised concerns about how to govern frontier models. Against that backdrop, member states met on November 19 to begin planning the second Global Dialogue on AI Governance. The permanent representatives of the Republic of Korea and Saint Kitts and Nevis, who co-chair the Dialogue, led the consultations. The Dialogue was created under General Assembly resolution 79/325 and will meet in New York on May 3–4, 2027. Estonia and El Salvador co-chaired the first Dialogue in Geneva in July. It noted that AI companies have since raised their own concerns about frontier models and human oversight.
Delegations discussed how the second Dialogue could build on the first and move from discussion toward practical implementation. Australia urged keeping two features that had worked in Geneva: the four thematic clusters and the multistakeholder format. The Philippines and Israel warned against merely repeating the debates of the first Dialogue. Estonia cautioned that governance conclusions agreed upon in Geneva were already being reopened in the Second Committee, urging against this practice. States also drew lessons from Geneva. The United Kingdom communicated their concerns about uncertainty over participation, which they considered unhelpful, and called for the dialogue to host real exchange rather than the mere reading off of set statements. Australia called for early clarity on the agenda and early circulation of the scientific panel’s report.
The European Union and developing countries stressed that AI cannot deepen inequalities nor leave less developed countries behind. Uruguay, for the Group of 77 and China, said developing countries should be helped to build their own AI systems. Guatemala and Iraq specifically called for the following: technology transfer, training, financing, infrastructure, and multilingual models.
They said these would let countries develop AI suited to their national needs rather than only use models built elsewhere. Andorra, speaking for the francophone ambassadors, said that language preconditions accessibility to AI.
India, Italy, Norway and Germany pushed for human-centered AI, human oversight and human rights across the debates. France, Belgium, Australia and Brazil singled out the protection of children.