Anthropic has opened a Seoul office. With the launch coinciding with U.S. government artificial intelligence export controls, it is being tested as a coordination channel between the company's headquarters and the South Korean government. The company said it aims to win in enterprise sales by putting forward trustworthy AI.

Chris Chaoui, Anthropic's head of international, said at a press briefing at the Conrad Hotel in Yeouido on June 17 that he did not think the export controls would remain in place. He said they appeared likely to be resolved within days.

On June 12 local time, Anthropic was hit with U.S. export controls that completely barred foreign nationals from accessing its latest models, Fable5 and Mythos5, for national security reasons. Work to grant access to South Korean companies and institutions, including the Korea Internet & Security Agency (KISA), Samsung Electronics, SK Hynix and SK Telecom, was also halted after KISA decided to participate in the second phase of Project Glasswing, a Mythos-based cybersecurity consultative body.

The issue came up repeatedly at the briefing. Chaoui stressed that the narrow scope of jailbreak mentioned this time was at a level that could commonly appear in models released over the past six months, and that he was confident the model would become usable again in Glasswing's second stage in the near term. In a follow-up question, he said to refer to the company's official position on its blog.

Anthropic's Seoul office is its third base in Asia, after Tokyo and Bengaluru. Chaoui said Anthropic, as an AI research company, pursues innovation with safety assured. He said South Korea, as one of a few countries worldwide to adopt comprehensive AI legislation, is an important market because it shares Anthropic's philosophy, explaining the background to the opening.

The organisation will be built around four major divisions under country head Ki-young Choi (최기영): sales, technology, policy and operations. Choi, a 30-year IT industry veteran who has worked at Microsoft, Autodesk, Adobe, Google Cloud and Snowflake, will lead efforts to secure enterprise customers. Support for developers and startups that has shown growth potential in the South Korean market will continue. Some non-profit and research projects tied to its public benefit corporation identity will also be pursued.

Ki-young Choi (최기영), Anthropic's South Korea head, said he had worked at IT companies for the past 30 years and experienced multiple technology transitions, and that this AI transition has the biggest impact above all. He said he would work to grow together with enterprises, organisations and research institutions in South Korea. Chaoui said the company would rapidly expand its business in South Korea within months.

It is also strengthening partnerships with cloud service providers, with enterprise sales in mind. Anthropic has partnerships with AWS Bedrock and Google Cloud Vertex AI, major global cloud service providers. It plans to expand this to Microsoft Azure and broaden cooperation to an AI infrastructure ecosystem that includes system integration.

It has also secured domestic system integration cooperation cases, including with LG CNS. The company said it would disclose additional partners as results become visible. Choi said it is important to create tangible business results together with system integration partners in terms of a partner ecosystem.

It is also reviewing building domestic infrastructure and data residency options to provide services suited to the corporate regulatory environment.

At the site, Choi introduced Naver, Hanwha Solutions and Nexon as major customer results. Ahead of the briefing, Anthropic Korea invited officials from major customers including Samsung Electronics, SK Hynix, SK Telecom and GS to explain its future direction.

It is also moving into long-term research cooperation. Working with the National AI Research Leading Lab (NAIRL), which includes KAIST, Korea University, Yonsei University and POSTECH, it will provide Claude free of charge to up to 60 researchers. It will support research on AI safety and model evaluation and analysis, and plans to continue expanding similar cooperation.

Government cooperation is also being made more concrete. Anthropic Chief Executive Dario Amodei and Bae Kyung-hoon met in February this year on the sidelines of the AI Impact Summit in New Delhi, India. The two sides decided to pursue official cooperation on the occasion of the Seoul office opening. Choi will directly handle government communications. There is not yet a separate policy and external affairs officer in the Seoul office. The headquarters policy team and a policy officer based in Tokyo, Japan, are supporting external affairs work in South Korea.