South Korea’s central bank, Naver launch sovereign AI system

The Bank of Korea (BOK) and Naver have launched a sovereign AI system called BOKI, marking the first such initiative by a central bank globally.

Developed over 18 months, BOKI operates entirely within the BOK’s internal network using cloud infrastructure and Naver’s HyperCLOVA X model.

Built by standardizing more than 1.4 million documents, the system aims to strengthen financial and economic data security and reduce reliance on external providers.

The on-premise setup challenges Korea’s traditional network separation rules and supports the country’s push for independent AI in the public and financial sectors, with a focus on data sovereignty.

BOK Governor Rhee Chang-yong said the project is intended as a model for public-sector digital innovation, while Naver founder Lee Hae-jin emphasized trust, stability, and the value of data assets.

🔗 Source: The Korea Times

🧠 Food for thought

Implications, context, and why it matters.

The broader impact of BOKI depends on whether Korea’s network separation rules are revised beyond this pilot

The Bank of Korea’s new AI system, BOKI, is being run as a pilot and test case and is being treated as an exception to Korea’s network separation rules, which require financial entities’ internal systems to be isolated from external networks.1
No details have been provided on a formal, sector-wide change to those network separation rules, including a timeline or which government bodies would be put in charge.1
With those gaps, it is hard to tell whether BOKI is being kept as a central bank only carve-out or is being used to open the door to wider cloud and AI use across public institutions and financial firms.1

A potential policy shift could expand demand for secure, on-premise AI deployments in the public sector

Cloud and AI vendors may see a larger market if the BOKI pilot leads to looser network separation rules, though that outcome is not confirmed.1
The central bank says it wants BOKI to serve as a model case and a “benchmark for public-sector digital innovation,” and it plans to share its experience and know-how with other public entities.1
That approach may raise interest in providers that handle compliance, security, and connecting AI models to legacy government IT systems, plus machine learning operations (MLOps) platforms, tools for deploying, monitoring, and governing AI in production, while no tenders or procurement plans are mentioned.1