Seoul Declaration

An international declaration for the safety, innovation, and inclusiveness of artificial intelligence, adopted at the 2024 AI Seoul Summit.

Seoul Declaration

Overview

The Seoul Declaration is an international declaration on the safety, innovation, and inclusiveness of artificial intelligence (AI), adopted at the AI Seoul Summit held in Seoul, Republic of Korea, on May 21, 2024. As a follow-up discussion to the 2023 Bletchley Declaration led by the United Kingdom, it aims to establish common safety principles and an international cooperation framework among countries in response to the rapid development of AI. The Republic of Korea and the United Kingdom served as co-chairs, and heads of state and minister-level figures from about 30 countries, along with representatives of major AI companies and international organizations, participated.

Main Contents

Background

The Bletchley Declaration, adopted at the first AI Safety Summit held at Bletchley Park in the United Kingdom in November 2023, was the first document to share international awareness of the potential risks of "frontier AI." Subsequently, in March 2024, the UN General Assembly unanimously adopted the first resolution on AI, which increased the need for follow-up discussions, and the Republic of Korea took on the role of co-host of the next conference.

Three Core Principles

The Seoul Declaration presents Safety, Innovation, and Inclusivity as the three pillars of AI governance.

1. Safety: Strengthen international cooperation for identifying, evaluating, and monitoring the risks of advanced AI models. Build a network among AI safety research institutes.

2. Innovation: Promote AI technology development and investment to the extent that it does not undermine safety, and create an open ecosystem for startups and researchers.

3. Inclusivity: Reduce the AI access gap and support capacity building and talent development in developing countries. Minimize discriminatory impacts based on gender, region, and class.

Seoul Statement of Intent

Separately from the summit declaration, the "Seoul Statement of Intent on International Cooperation for AI Safety Science," signed by representatives of national AI safety research institutes, was announced. This is an inter-institutional cooperation framework for joint research and information sharing on AI safety evaluation methodologies, risk thresholds, red-team testing, and the like. The United States, the United Kingdom, Japan, France, Germany, Canada, Australia, Singapore, and others participated, and it later developed into the "International Network of AI Safety Institutes."

Voluntary Corporate Commitments

Major global AI companies also announced voluntary safety commitments in line with the declaration. These included commitments by frontier model developers to disclose risk management frameworks, conduct safety evaluations before releasing models, and transparently disclose incidents when they occur.

Major Participating Countries and Organizations

About 30 countries, including the Republic of Korea, the United Kingdom, the United States, Japan, France, Germany, Italy, Canada, Australia, Singapore, India, and Brazil, as well as international organizations such as the United Nations (UN), the Organisation for Economic Co-operation and Development (OECD), and the European Union (EU), participated. Among technology companies, representatives of major big tech firms and AI startups attended.

Follow-up Measures

After the Seoul Declaration, the third AI Action Summit was held in Paris, France, in 2025, and countries continued cooperation among AI safety institutes, regulatory sandboxes, and the introduction of AI impact assessment systems. The key achievement of the Seoul Summit lies in presenting a balanced framework that combined the "safety" discourse with "innovation" and "inclusivity," and in creating practical channels of cooperation among institutions.

Latest Trends

In 2024–2025, follow-up discussions to the Seoul Declaration are unfolding in three directions. First, the International Network of AI Safety Institutes has begun substantive joint research (model evaluation standards, sharing of risk scenarios). Second, as countries enact AI basic laws and regulations, work is underway to convert the declaration's principles into domestic law, with the implementation of the EU AI Act and Korea's discussions on an AI basic law being representative examples. Third, as AI safety becomes combined with military and security issues, "AI governance" has emerged as a diplomatic agenda, and follow-up norm discussions continue in multilateral frameworks such as the UN and the OECD. However, limitations are also pointed out, such as the absence of binding international norms, the technology gap between developed and developing countries, and security competition.