Japan and South Korea Launch Joint AI Safety Institute

Aisafety — Japan and South Korea have joined forces to create a Joint AI Safety Institute, which aims to standardize AI safety evaluation frameworks across both countries starting in June 2026.

This collaboration marks a significant step toward aligning regional standards for testing large language models (LLMs) and managing risks. While the official name for this joint body is still pending, this partnership shows a shared dedication to building secure and interoperable AI governance in East Asia.

This initiative comes on the heels of South Korea launching its own AI Safety Institute in November 2024, a decision that arose from commitments made at the AI Safety Summit in Seoul in May 2024.

Japan has also been proactive, having established the Japanese AI Safety Institute (AISI) in April 2024 via its Ministry of Economy, Trade and Industry (METI). Japan’s AISI has already made strides by conducting its first LLM safety evaluations in 2024 and publishing performance findings on models from various international developers. You can find more details in a recent article on the OpenAI Blog.

The real takeaway here isn’t just the announcement; it’s about the technical integration necessary for these frameworks to work across borders. Both nations are already members of the International Network of AI Safety Institutes, which had its inaugural meeting in San Francisco in November 2024, as reported by the VentureBeat AI.

This existing membership likely lays the groundwork for their bilateral collaboration. As of June 9, 2026, though specifics about joint technical standards, funding, or staffing haven’t been confirmed, this move indicates a shift from fragmented national oversight to a more cohesive, cross-border regulatory framework.

Not everyone is on board with the pace of this integration. Some industry analysts believe national sovereignty concerns over proprietary model data could slow down the standardization process. However, the push for a common understanding in AI safety is motivated by the pressing need to tackle systemic risks that cross borders. With the global AI market expanding rapidly, Japan and South Korea are positioning themselves as key players in creating a stable, predictable, and secure AI ecosystem.

What’s still unclear is the specific focus areas for model auditing that the joint institute will prioritize. Future updates should reveal whether the institute will concentrate on red-teaming, bias detection, energy efficiency, or perhaps all of these aspects. For the moment, the industry is keeping a close eye on how this partnership might influence the broader Global AI Alliance research initiatives that are shaping international policy.

FAQs

Aisafety: When was the Japanese AI Safety Institute established?

Japan’s AI Safety Institute was formally established in April 2024 under the direction of the Ministry of Economy, Trade and Industry (METI).

Aisafety: What was the significance of the May 2024 AI Seoul Summit?

The AI Seoul Summit, co-hosted by South Korea and the United Kingdom, served as the catalyst for South Korea to launch its national AI Safety Institute in November 2024.

Are these nations part of a larger global safety network?

Yes, both Japan and South Korea are members of the International Network of AI Safety Institutes, which held its inaugural meeting in San Francisco in November 2024.

Why are Japan and South Korea establishing a Joint AI Safety Institute?

Japan and South Korea are establishing the Joint AI Safety Institute to create standardized evaluation frameworks that ensure the responsible development and deployment of artificial intelligence technologies across both nations.

When will the Joint AI Safety Institute begin its operations?

The specific launch date and operational commencement of the Joint AI Safety Institute have not been independently confirmed as of June 9, 2026.