ARTEX AI Hacking Tool Breaches South Korean Banks

According to a report by CrowdStrike covered by The Decoder, a suspected single attacker successfully breached multiple South Korean financial institutions between late September and early October 2026. The attacks resulted in the theft of over 25,000 customer records containing names, contact details, income data, and credit limits at Shinhan Bank alone.

How did a single attacker execute massive bank breaches?

The operation relied on ARTEX, an open-source penetration testing tool hosted on GitHub that automates vulnerability discovery using large language models. The technical logs revealed that the attacker leveraged models including DeepSeek v4.1-flash, GLM-5.3, and Grok 4.6 to scan and exploit network defenses without requiring a large coordinating team.

What are the practical consequences for cybersecurity teams?

This incident demonstrates that generative AI tools drastically lower the technical barrier and workforce requirements needed to execute enterprise-scale data breaches. Security teams must now defend against automated threats capable of finding flaws and executing exploits at machine speed, requiring a shift toward automated defensive postures and tighter monitoring of open-source security utilities.