AI Safety Institute (Korea AISI)
- Type
- Government body
- Lieu
- Corée du Sud — Republic of Korea
- Nom local
- 인공지능안전연구소
- Dernière vérification
- 2026-09-22
- Prochaine vérification
- 2027-03-21
Pas encore traduit — affiché en anglais.
Sièges nominatifs
- KIM Myuhngjoo — chair
Comment les joindre
- Korea AISI English site
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.
- Korea AISI 'Collaboration' page
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.
- Korean-language site
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
- Public submission route
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.
Ce qu'il fait
Launched Nov 2024. Hosted within ETRI (한국전자통신연구원, Electronics and Telecommunications Research Institute) under the Ministry of Science and ICT (과학기술정보통신부, MSIT). Three divisions: AI safety policy, AI safety evaluation, AI safety research. Supports MSIT in implementing the AI Framework Act but is itself a research institute — it does not issue orders or fines.
Ses responsabilités en matière d'IA
Statutory body established under the AI Framework Act to evaluate AI risk and develop safety standards. Korea's node in the International Network of AI Safety Institutes. Co-produced the bilingual Korea AI Basic Act resource with the Future of Life Institute.
Ses pouvoirs
research
Évaluation franche
Poor. Korea AISI has an English site that advertises international collaboration but publishes no contact email, no form, and no submission route — verified by fetching both the English homepage and the dedicated Collaboration page. The only published contact is a street address in Pangyo. For a foreigner this is effectively unreachable except via a named researcher or through ETRI/MSIT.
Évaluation franche
Not a participation route in any formal sense, but it is the most approachable Korean AI institution for an outsider because it has a working English presence and an internationally-facing research remit. Researchers and technical contributors can realistically make contact. Ordinary citizens have no route here and should use the petition or the MSIT pre-announcement instead.
Travaux en cours
Operates under MSIT with the Electronics and Telecommunications Research Institute (ETRI) as host/supporting organisation — note this dual line: it is statutorily grounded but institutionally embedded in a government research institute rather than standing alone. Work programme covers AI safety policy, evaluation methodology and international AISI collaboration.
Voie d'accès publique
No formal comment mechanism. Practical routes: research collaboration enquiries and general contact via aisi.re.kr; international engagement through the AI Safety Institute network; policy input indirectly via MSIT pre-announcement.
Règles de format
None; no formal participation process exists to have a threshold.
Langues
Korean and English. Korea AISI maintains a genuine English site — a notable exception in this jurisdiction and the most accessible Korean AI body for a non-Korean speaker.
Notes
I could not verify any 2025–26 rename. Some sources render the English name as 'Korea AI Safety Institute'; the site itself says simply 'AI Safety Institute (AISI)'.
Note de recherche
Director and the MSIT/ETRI institutional relationship VERIFIED from the institute's own English site aisi.re.kr/eng/contents/23 and its site footer. The statutory basis under the AI Framework Act is VERIFIED indirectly via Cooley's analysis, which refers to an 'AI Safety Research Institute' established by the law; the exact article number is UNVERIFIED.