Resource guide

UN AI Scientific Panel and Global Dialogue Explained

Their founding resolution gives the panel assessment work and the Dialogue discussion work, with no rulemaking tasks.

Last updated November 17, 2026 1574-word guide By Jonathan West

What the UN Created for AI Governance

The United Nations (UN) General Assembly created two artificial intelligence (AI) bodies on 26 August 2025, in resolution A/RES/79/325. One is a scientific panel that assesses research. The other is a forum for discussions among governments and stakeholders. Their assigned work explains what UN AI governance does, and where its limits lie.

Resolution A/RES/79/325 establishes the Independent International Scientific Panel on Artificial Intelligence in paragraph 1 and the Global Dialogue on Artificial Intelligence Governance in paragraph 4. The Panel assesses existing research on AI's opportunities, risks and impacts. The Dialogue brings governments and other stakeholders together to discuss cooperation and share practices.

Both bodies' pages were retrieved November 17, 2026. A scientific assessment can inform a debate, and a discussion can bring participants together. Neither task is an instruction to inspect an AI system or enforce a requirement.

The resolution's preamble recalls the Global Digital Compact, an annex to the Pact for the Future. That earlier agreement called for a scientific panel and a global dialogue. Resolution A/RES/79/325 sets out how the two bodies are to function. The terms of reference describe two bodies. They do not create a single worldwide rulebook.

The Two Bodies at a Glance

The Panel supplies assessments. The Dialogue supplies a place to discuss AI governance. The resolution connects them by directing the Panel to present its annual report at the Dialogue.

BodyMembersWhat it doesWhat it producesResolution paragraph
Scientific Panel40 members serving in their personal capacitySynthesizes and analyzes existing research on AI opportunities, risks and impactsOne annual policy-relevant but non-prescriptive summary report, with thematic briefs as it deems necessary1 and 1(a)
Global DialogueGovernments and relevant stakeholders take partDiscusses international cooperation and shares best practices and lessons learnedAnnual discussions and a Co-Chairs' summary4, 5 and 7

The table's "members" column describes different arrangements. Paragraph 1 specifies a fixed membership for the Panel. Paragraph 4 instead describes who the Dialogue involves, and paragraph 5 describes its meeting format. A Dialogue participant does not thereby become a Panel member.

What the Scientific Panel Does

The Panel's central assignment is to synthesize and analyze existing AI research, according to resolution paragraph 1(a). Its assessments cover opportunities, risks and impacts. The same paragraph describes the annual output as "policy-relevant but non-prescriptive": the report can address questions relevant to policy without prescribing rules that governments must adopt.

Paragraph 1 establishes a multidisciplinary Panel with balanced geographic representation. Its 40 members serve in their personal capacity for three-year terms. Paragraph 3 provides for a published, criteria-based open call and a list of recommended candidates from the UN Secretary-General for appointment by the General Assembly. It also sets limits on shared nationality or affiliation and excludes employees of the UN system.

Members serve in their personal capacity under paragraph 1. The resolution does not describe the annual report as a negotiated statement by governments. Paragraph 2 requires candidates to disclose financial, professional and personal conflicts of interest. Appointed members must continue disclosing conflicts as they arise.

The Panel can organize working groups and consult external experts informally when needed, under paragraph 1(c). Paragraph 1(b) directs members to elect two Co-Chairs, one from a developed country and one from a developing country, along with up to three Vice-Chairs. Under paragraph 1(d), the Co-Chairs can provide updates through an interactive dialogue with the General Assembly plenary up to twice a year.

The annual report has another destination. Paragraph 1(e) directs the Panel to present it at the Global Dialogue. A claim attributed to the Panel should distinguish its report, a thematic brief and an update from its Co-Chairs. The resolution describes those outputs in different terms.

What the Global Dialogue Does

The Global Dialogue is a platform for discussion among governments and relevant stakeholders, as resolution paragraph 4 states. Its stated purposes include discussing international cooperation, sharing best practices and lessons learned, and holding open, transparent and inclusive conversations about AI governance.

Paragraph 4 sets out subjects those conversations can cover. They include safe, secure and trustworthy AI systems; gaps in the ability to work with AI; and AI's social, economic, ethical, cultural, linguistic and technical implications. The paragraph also names human rights, transparency, accountability and human oversight consistent with international law. These are discussion topics in the resolution. The resolution does not describe them as requirements for an AI developer.

The Dialogue also addresses how governance approaches can work together, along with open-source software, open data and open AI models. Paragraph 4 connects its discussions to the Sustainable Development Goals and closing digital divides. A country seeking to share an approach and a stakeholder raising a capacity gap can therefore participate in the same forum without the resolution treating their contributions as binding rules.

Under paragraph 5, the Dialogue convenes annually for up to two days in the margins of existing UN conferences, alternating between Geneva and New York. Its format includes a multi-stakeholder plenary, a high-level governmental segment, presentation of the Panel's annual report and thematic discussions. Under paragraph 7, the President of the General Assembly is invited to appoint two Co-Chairs for each Dialogue. The Co-Chairs identify main topics and draft a Co-Chairs' summary.

Paragraph 8 gives the Co-Chairs of the second Dialogue a further job: holding intergovernmental consultations on common understandings in priority areas of international AI governance. Those understandings are intended as input to the high-level review of the Global Digital Compact. Paragraph 8 does not describe those consultations as an enforcement process.

What Neither Body Is Set Up to Do

The resolution lists assessment and reporting for the Panel, and discussion and consultation for the Dialogue. It lists no rulemaking, inspection or enforcement tasks for either body in paragraphs 1 and 4 through 8. This is a reading of their assigned functions in A/RES/79/325. It does not settle what governments might agree to through a different process.

The preamble draws a separate boundary: the resolution and the activities of both bodies are limited to the non-military domain and do not refer to AI for military purposes. A question about weapons belongs on a different policy track. The autonomous weapons treaty push explainer covers that subject. For rules made within countries, start with the separate discussion of national AI laws.

A UN forum can discuss human oversight, and a UN scientific body can assess risks, without either action itself creating an enforceable requirement. To evaluate a specific requirement, identify the instrument that states it and the authority responsible for it. The founding resolution of the Panel and the Dialogue supplies their remit and lists no inspection procedure for AI products.

What the Panel and Dialogue Have Produced So Far

The UN Panel page listed a "Preliminary Report" and a "Thematic Brief on AI Agents, Misalignment and the Risk of Losing Human Control" as of November 17, 2026. The page listed them with 01 July and 21 September dates, respectively. Their titles establish that the items were listed. Neither publication's findings are described here.

The UN Dialogue page reported, as of November 17, 2026, that the inaugural Dialogue took place in Geneva on 6 and 7 July 2026. It reported delegations from 163 countries, alongside participants from the private sector, civil society, academia and the technical community. The same page said the Panel presented its Preliminary Report at the Geneva Dialogue on 6 July 2026 and listed a published Co-Chairs' Summary. None of those facts makes the summary a rule.

As of November 17, 2026, the UN Dialogue page scheduled the next Dialogue for 3 to 4 May 2027 in New York and said the Panel's next annual report would inform it. That schedule describes the planned forum. It does not describe the eventual content of its discussions. The EU AI Act provides a different kind of comparison: it is a binding regional law with penalties described in the site's separate explainer. The UN bodies' listed outputs are assessments and discussions.

Our Takeaway

The resolution creates a way to gather scientific assessments and discuss AI governance across countries. Paragraph 1 provides for a 40-member Panel and one annual non-prescriptive summary report. Paragraphs 4 and 5 provide for a recurring Dialogue among governments and stakeholders. The preamble limits both to the non-military domain. For a reader asking whether these bodies can restrain an AI system through rules or enforcement, their listed tasks in A/RES/79/325 do not provide that mechanism.

The Panel presents its annual report at the Dialogue under paragraph 1(e). Governments can consider research presented by the Panel and ideas discussed at the Dialogue. Whether those ideas later become binding obligations depends on a separate decision and its terms. The resolution's own wording supports an account of evidence and discussion, It does not show that these bodies enforce worldwide AI rules.

What Would Change Our Answer

A change to the bodies' terms of reference could change this reading. Resolution paragraph 12 says the General Assembly may consider and decide whether to continue those terms during the Global Digital Compact's high-level review at its eighty-second session. Any decision that changes the Panel's or Dialogue's assigned tasks would require a fresh reading of the adopted text.

A binding instrument adopted by member states would also require its own assessment. Its obligations would come from that instrument's wording. To read the assignments, open resolution A/RES/79/325, starting with its preamble and paragraphs 1, 4 and 5.

Frequently asked questions

▸ What is the UN AI Scientific Panel?
It is a scientific assessment body established by the United Nations General Assembly in resolution A/RES/79/325. Paragraph 1 provides for 40 members serving in their personal capacity. Its assigned work includes synthesizing existing research and issuing one annual policy-relevant but non-prescriptive summary report.
▸ What is the Global Dialogue on AI Governance?
It is a forum for governments and other stakeholders to discuss international cooperation, share practices and consider AI governance, as described in paragraph 4 of resolution A/RES/79/325. Paragraph 5 provides for an annual gathering of up to two days.
▸ Is UN AI governance binding?
The resolution gives these two bodies assessment, reporting and discussion tasks. It does not list rulemaking, inspection or enforcement among those tasks. That is a reading of these bodies' assigned functions. It makes no claim about other forms of international AI cooperation.
▸ When is the next Global Dialogue on AI Governance?
The next Dialogue is scheduled for 3 to 4 May 2027 in New York, according to the United Nations Dialogue page as of November 17, 2026. The schedule may change.