Partnership on AI (PAI)
- Type
- Civil society organisation
- Lieu
- International — International (US-based non-profit)
- Dernière vérification
- 2026-09-22
- Prochaine vérification
- 2027-03-21
Pas encore traduit — affiché en anglais.
Comment les joindre
- Contact form
- Become a partner organisation
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.
- Homepage
Ce qu'il fait
Multistakeholder convener with 100+ partner organisations spanning Amazon, Google, Meta, Microsoft, OpenAI, Apple alongside ACLU, ARTICLE 19, Future of Privacy Forum, UNICEF, UNDP and universities. Produces guidance documents and frameworks — notably on safe foundation model deployment, synthetic media, and AI and labour — across four programme areas: AI and Human Connection; AI, Labor and the Economy; Safety-Critical AI; and Public Policy.
Évaluation franche
Consensus-bound. With the largest labs as partners, PAI's outputs are negotiated to the level everyone can accept, which limits how sharp a concern it will carry. Organisational partnership is possible; individual input is not really a thing. Middling responsiveness.
Préoccupations couvertes
Deliberately plural. Its Safety-Critical AI programme touches frontier deployment risk; its labour and human-connection work is squarely near-term/structural. Its distinctive contribution is process norms rather than a risk thesis.
Canal gouvernemental
Moderate — PAI guidance is cited in policy processes and it participates in multistakeholder forums, but it is a norm-setter, not an advocate.