South Korea's Ministry of Science and ICT said on Wednesday it signed a memorandum of understanding with global artificial intelligence (AI) research and development company Anthropic to cooperate on AI safety and cybersecurity.

The ministry said the MOU will be used to analyse AI's impact on cyber attacks and defence. It will also assess AI model safety and misuse risks in a Korean-language context. The ministry said it will also strengthen a cooperation system to secure safety, including red-team evaluations of autonomous AI agents.

The MOU is a follow-up measure to cooperation plans discussed when Vice Prime Minister and Science Minister Bae Kyung-hoon (배경훈) met Anthropic Chief Executive Dario Amodei at the 2026 India AI Impact Summit in February.

Anthropic officially opened its South Korea office the previous day. It is the fourth after Japan, India and Australia.

The government has built a cooperation system with four global AI companies, including Anthropic, following Nvidia, OpenAI and Google DeepMind.

After the MOU signing, Lee Do-gyu (이도규), director general for information and communications policy at the ministry, met Anthropic global lead Chris Chaouli to discuss the direction of cooperation on AI safety and cybersecurity.

In AI safety, the Korea AI Safety Institute (AISI) and Anthropic agreed to cooperate on safety evaluations of AI models and autonomous AI agents. In cybersecurity, they agreed to cooperate on finding AI vulnerabilities, including in the financial sector, and sharing expertise and information related to cyber threats.

Lee said, "The next 2 to 3 years in the global AI market will be a huge showdown that will decide hegemony and a golden time," and added, "Cooperation with Anthropic will be a strong driving force for South Korea's AI innovation."

Chaouli said, "South Korea is one of Anthropic's most important markets in Asia-Pacific," and added, "Cooperating with government agencies that share the value of developing safe and responsible AI is core to Anthropic's operations."