The South Korean government has proposed a collaboration with the American AI company Anthropic, focusing on AI safety and cybersecurity through the AI Safety Research Institute (AISI).
According to the Ministry of Science and ICT, a meeting took place on the morning of May 11 in the Gwanghwamun conference room in Seoul, attended by Deputy Minister Ryu Je-myung, AISI Director Kim Myung-joo, Korea Internet & Security Agency (KISA) Director Oh Jin-young, and Anthropic's Global Policy Lead Michael Selitto. Representatives from the Ministry of Foreign Affairs, the National Intelligence Service, the Financial Services Commission, and the Financial Security Institute were also present.
This meeting was organized as a follow-up to discussions between Deputy Prime Minister and Minister of Science and ICT Bak Hoon and Anthropic CEO Dario Amodei during the '2026 India AI Impact Summit' held in February. At that time, both parties discussed the influence of agent AI, AI safety and security collaboration, and strengthening global leadership.
The Ministry proposed establishing a collaboration framework centered on AISI for verifying the safety of AI models and enhancing cybersecurity. They emphasized the need for a system to share information that would allow South Korea to proactively respond to global vulnerabilities and security patches. Both sides agreed to continue practical discussions regarding the cybersecurity applications of AI models.
Additionally, discussions were held regarding the domestic AI regulatory environment, including the AI Basic Act and AI safety policy directions. A ministry official stated, "There were discussions on AI safety policies, the direction of the AI Basic Act, and global AI security collaboration systems," adding that Anthropic showed significant interest in the South Korean government's AI policies and regulatory framework.
Deputy Minister Ryu remarked, "As the performance and application scope of frontier-level AI models rapidly expand, it is crucial to create an environment where citizens and businesses can safely utilize AI. We will actively collaborate with leading global AI companies to enhance the safety of AI models and cybersecurity capabilities."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.