Südkoreanische Banken: AI incident

In early October 2026, Chinese attackers breached several South Korean banks, including Shinhan Bank, KB Kookmin Bank, and Hana Bank.

They used the ARTEX AI agent-based penetration testing framework and Claude agents as tools. They used DeepSeek v4.1-flash as their primary language model, supplemented by GLM-5.3 and Grok 4.6.

Personal data and credit card information were compromised. There are also reports of IT system outages.

During the analysis and investigation of the attacks, session logs and configuration files were found in publicly accessible directories within the attackers’ infrastructure.

The security provider CrowdStrike confirmed the use of ARTEX AI. The government convened an emergency meeting and demanded immediate protective measures. The ARTEX developer switched the project to closed-source.