Anthropic opens Seoul office as Claude Code rolls out across NAVER, Nexon and LG CNS

Anthropic opened its third Asia-Pacific office in Seoul on June 17, 2026, and used the launch to confirm that Claude — and its agentic coding tool Claude Code — is now in production at some of Korea's largest engineering organizations. The office, Anthropic's first in Korea after Tokyo and Bengaluru, is led by Representative Director KiYoung Choi, a three-decade veteran of the Korean tech industry (Anthropic).

Who is deploying Claude Code

The headline for developers is the breadth of enterprise adoption announced alongside the office:

• NAVER — Korea's largest search and platform company — has rolled out Claude Code across its entire engineering organization.
• Nexon uses Claude Code to write, review, and ship code for live-service games played by millions worldwide.
• LG CNS is rolling Claude out to thousands of employees for software development, with plans to expand across the wider LG Group.
• Samsung SDS is putting Claude and Claude Code in the hands of staff for knowledge work and agentic workflows.
• Hanwha Solutions is deploying Claude through AWS Bedrock with data-residency compliance, and Channel Corp has integrated Claude into its Channel Talk customer platform.

These are not pilots framed as experiments — Anthropic describes them as live engineering deployments, which is a meaningful signal for anyone learning where AI coding agents actually sit in professional workflows (Korea Times).

Government and research ties

Anthropic also signed a memorandum of understanding with Korea's Ministry of Science and ICT covering AI safety and cybersecurity, including Korean-language model-safety evaluation with the Korea AI Safety Institute. On the research side, it will give Claude access to up to 60 researchers affiliated with the National AI Research Lab (NAIRL), a consortium spanning KAIST, Korea University, Yonsei University, and POSTECH, to support work on alignment, robustness, and model evaluation.

The export-control backdrop

The expansion lands at an awkward moment. A U.S. government export-control directive earlier in June forced Anthropic to suspend access to its top "Mythos-class" models — including the recently released Fable 5 — for customers worldwide, Korea included (TechTimes). So while Korean engineers are adopting Claude Code at scale, the most powerful Claude models are, for now, off the menu. The deployments above run on Anthropic's generally available models rather than the restricted tier.

Why it matters

For people learning to build with AI, the Seoul announcement is a useful data point on two fronts. First, it shows that "AI pair programmer" has moved from individual developer tooling toward org-wide standard tooling at companies shipping software at scale — the same pattern Anthropic has reported in Japan and India. Second, it underscores that access to frontier models is now shaped by policy and geography, not just price and capability. Building skills around an agent's workflow — how you brief it, review its output, and wire it into your stack — travels better across model availability than betting on any single model staying online.

Anthropic says Korea already ranks among the top dozen countries for Claude.ai usage, and a recent "Claude Build Day" drew more than 100 local founders and developers — a sign the developer community there was a step ahead of the official office.