Third revision of draft UNGA resolution for Scientific Panel on AI and Dialogue on AI Governance

Flags at United Nations Headquarters - New York, USA

When UN member states adopted a Global Digital Compact in September 2024, they decided to establish a Scientific Panel on AI and a Global Dialogue on AI Governance. A process to define the terms of reference and modalities for the Panel and the Dialogue has been unfolding in New York since January 2025, with member states negotiating over a draft resolution.

On 24 June 2025, a third revision (Rev3) of this draft resolution was published, bringing a number of key changes and clarifications compared to the second revision (Rev2).

One of the additions in Rev3 is an explicit clarification that the resolution addresses AI in the non-military domain only, making clear that it does not apply to military uses of AI.

The text further strengthens governance provisions for the Scientific Panel. New elements include:

  • The possibility of forming working groups within the Panel and the consultation of external experts, which was not present in Rev2.
  • More frequent reporting by the Panel to the General Assembly: up to twice per year instead of once annually.
  • A ban on UN employees serving on the Panel and a limit of two members per nationality, adding sharper rules on representation and independence.
  • Stronger language on conflict of interest disclosures, now framed as a requirement.

Regarding the Global Dialogue on AI Governance, the third revision:

  • Introduces new thematic areas for discussion, including open-source software, open data, and open AI models, expanding on the dialogue’s scope.
  • Adds a rotational hosting format alternating between Geneva and New York, compared to Rev2 which did not specify this.
  • Calls for the appointment of co-facilitators to prepare a separate resolution on AI governance prior to the 2027 High-Level Review of the Global Digital Compact, a task not included in Rev2.
  • Enhances the role of the broader UN system by specifying that agencies, programmes, and inter-agency mechanisms should actively contribute to international AI capacity-building efforts.

Other refinements include greater emphasis on transparency in funding and more precise language around the continuation and review of the Panel and Dialogue.