Verified, then logged.
Sun 22 Nov 2026 · 00:34 KST / 15:34 UTC한국어 | ENSearch /

Breaking04 / Policy

Korea AI Safety Institute sets workshop on AI risk practices

A video camera on a tripod faces a lectern and blank screen in an empty hotel conference room set with microphones and closed laptops. AI-generated image

The Korea AI Safety Institute, which handles Korea's cooperation with AI safety institutes and related bodies abroad, announced 23 Oct 2026 as the date of its sixth AI safety policy research seminar. The event is a forum for discussing a research agenda, not the passage or entry into force of any law.

The seminar is named the 2026 AI Safety Policy Global Workshop and has the theme "From Framework to Practice", according to the institute's notice. The venue is the Hotel Shilla in Seoul, with a parallel online format on Zoom. Seating for in-person attendance is limited and requires prior registration and approval.

The first talk of the opening session, according to the programme, presents the results of a Delphi study drawing on the views of 272 international experts; a Delphi study is a method that gathers expert opinion through repeated rounds of questions. The speaker is Michael Noetel, a senior researcher at the MIT AI Risk Initiative. Akira Yoshida, head of policy at Anthropic, is listed in the same session with a talk on how the company operates its risk framework.

The programme lists three speakers for the second session: Kyungwoo Song, an associate professor at Yonsei University, a private university in Seoul; Melissa Yoong, a director at Microsoft; and Jiyeon Cho, a senior researcher at the institute. Song's talk looks beyond benchmarks, the tests used to compare performance, to AI risk assessment and quantification in real-world use.

The institute describes the seminar as an annual series for discussing key issues in AI risk and safety policy and identifying a mid- to long-term research agenda. It said the sixth edition would "explore the practical application of AI safety frameworks and their policy implications".

The notice does not say whether the outcomes of the discussions will be made public, and it sets no timetable for follow-up policy.