Français

Organismes ayant un pouvoir

Independent International Scientific Panel on AI (UN)

Type
Multilateral body
Lieu
International — Global (UN General Assembly)
Dernière vérification
2026-09-22
Prochaine vérification
2027-03-21

Pas encore traduit — affiché en anglais.

Comment les joindre

  • Mailing list for Panel updates (only public-facing mechanism found) · Web form
    Vérification expirée · 2026-09-22

    Ce canal est conservé uniquement comme référence. Sa vérification est absente, date de plus de 180 jours, ou la période de contact n'est pas ouverte actuellement ; ne vous y fiez pas avant une nouvelle vérification.

    Notifications only — not a submission channel

    Linked from the Panel homepage footer.

  • Generic UN contact form (not Panel-specific) · Web form
    Vérification expirée · 2026-09-22

    Ce canal est conservé uniquement comme référence. Sa vérification est absente, date de plus de 180 jours, ou la période de contact n'est pas ouverte actuellement ; ne vous y fiez pas avant une nouvelle vérification.

    General UN correspondence

    No Panel-specific inbox exists; mail here is very unlikely to reach members.

  • Homepage · Homepage
    Non contrôlé

Ce qu'il fait

40-member panel created by UNGA resolution (Aug 2025), members appointed by the General Assembly 12 Feb 2026, co-chairs Yoshua Bengio (Canada) and Maria Ressa (Philippines) elected 3 Mar 2026. Produces one annual evidence-based assessment of AI opportunities/risks/impacts plus thematic briefs; may update the GA up to twice a year. Preliminary Report launched 1 July 2026; first full annual report due at the second Global Dialogue (New York, May 2027).

Évaluation franche

Effectively closed to outsiders. The FAQ says only that the Panel 'may consult informally with external experts' and may form working groups; there is no call for evidence, no submission portal, and no published contact address. The realistic route in is to reach an individual Panel member through their home institution, or to route through the Global Dialogue's written-inputs process instead.

Préoccupations couvertes

Deliberately spans the whole range — capability/risk assessment covering misuse, loss-of-control, and systemic/societal impacts — modelled on the IPCC. It is an assessment body, not an advocacy body, so it has no single threat model.

Canal gouvernemental

Maximal — reports directly to the UN General Assembly and to the Global Dialogue on AI Governance. But the channel runs one way, from Panel to states.

Notes

Members were selected from 2,600+ applicants across 140+ countries. Bengio simultaneously chairs the International AI Safety Report and runs LawZero, so he is a heavily saturated single point of contact.

Sources

Une erreur ?