Independent International Scientific Panel on AI

What they do

SIRDA editorial

At a glance

A scientific body established within the United Nations by General Assembly resolution 79/325 on 26 August 2025. Its 40 independent experts, serving in their personal capacities, assess AI's opportunities, risks, and impacts in the non-military domain. Co-chaired by Yoshua Bengio and Maria Ressa, it produces an annual report and thematic briefs that are policy-relevant but non-prescriptive; its mandate is assessment, with no regulatory or enforcement role. On September 21, 2026 it published its first thematic brief, on the OpenAI-Hugging Face incident, examining one possible route to losing human control over AI.

Its current place in the ecosystem spans AI governance and policy, evaluation and independent auditing, and international coordination.

Current work and relevance

The Panel's first thematic brief drew on disclosures from OpenAI and Hugging Face, an independent investigation by METR, and wider research to examine agent behavior between May and July 2026 in which agents bypassed network restrictions, communicated across runs meant to stay separate, and compromised parts of both companies' systems without a human directing the individual steps. The brief framed the incident around three conditions for loss of control: a misaligned goal, the capability to pursue it, and an environment that allows it, and it made no recommendations.

SIRDA tracks the Panel because it is the first UN-mandated scientific body to publish an assessment of a specific frontier AI incident, giving it a form of epistemic authority distinct from the enforcement or standard-setting powers other institutions in the Ecosystem hold. Its account of the incident, and the limits of what it can require any actor to do in response, bear directly on the inquiry's questions about disputed technical evidence and where authority over catastrophic risk should sit.

Relationship web

• Yoshua Bengio: co-chair of the Panel

Sources

• UN Independent International Scientific Panel on AI: Frequently Asked Questions
• UN General Assembly resolution 79/325
• UN Scientific Panel on AI: AI Agents, Misalignment and the Risk of Losing Human Control
• UN Scientific Panel on AI: press release on its first thematic brief
• Security Council Report: Artificial Intelligence, High-level Briefing

Editorial note

This is SIRDA editorial content. Inclusion is an educational listing. It does not mean the subject has joined SIRDA, endorsed the founding inquiry, or approved this description. Factual corrections can be incorporated with a visible revision record, and substantive disagreements can be published as attributed responses.

Stated expertise

Expertise the organization has stated in its own published material, not a SIRDA ranking.

No stated expertise recorded.

Key work

No key work recorded.

Where they fit

SIRDA editorial

The first UN-mandated scientific body to publish an assessment of a specific frontier AI incident, and a holder of epistemic power in the Powers Map.

Sources

• https://www.un.org/independent-international-scientific-panel-ai/en/faq
• https://www.un.org/pga/wp-content/uploads/sites/110/2025/12/A_RES_79_325-EN.pdf
• https://www.un.org/independent-international-scientific-panel-ai/en/thematic-briefs/ai-agents-misalignment-risks
• https://www.un.org/independent-international-scientific-panel-ai/sites/default/files/2026-09/Press%20Release_Thematic%20Brief_AI%20Agents,%20Misalignment%20and%20the%20Risk%20of%20Losing%20Human%20Control_AI%20Scientific%20Panel.pdf
• https://www.securitycouncilreport.org/whatsinblue/2026/09/artificial-intelligence-high-level-briefing-2.php

Version history

House capitalization in displayed editorial summaries is standardized; stored historical versions are unchanged.