Attackers using the ARTEX AI pentest agent stole customer data from South Korean banks
What happened
At least seven South Korean banks and lenders, including Shinhan, KB Kookmin and Hana, reported intrusions from late September 2026 through weakly protected side systems such as a loan recruiter lookup tool. Officials and CrowdStrike tie the attacks to ARTEX, an open-source agentic penetration testing tool, used with Claude Code and several commercial models. Names, phone numbers, incomes and loan limits were exposed.
The breaches came to light on 30 September 2026. Attackers targeted supporting systems rather than core banking: a lookup tool Shinhan built for outside loan recruiters, a mobile work-support system at KB Kookmin and a sales-support system at Hana. Shinhan put its exposure at about 25,000 customers; KB Kookmin reported 99 customers and 20 current and former employees, Hana 89 customers and BNK Busan 11 outsourced developers. Press totals range from about 65,000 to at least 68,000 people.
President Lee Jae Myung said signs had emerged that AI agents were used, and a Financial Security Institute official said logs pointed to ARTEX with human involvement. CrowdStrike found open directories on attacker servers exposing Claude Code session histories and ARTEX configuration linked to the attacks; ARTEX used DeepSeek v4.1-flash, and the actor also used GLM-5.3 and Grok 4.6. CrowdStrike assesses the actor as a financially motivated Chinese speaker, with moderate confidence.
How it happened
AI-run attack: Autonomous intrusion. The attacker's own AI agent carried out the break-in, working through the steps with little human involvement.
Models from several AI vendors were involved. The product was ARTEX, which is open source, a type of AI agent.
Impact
Data was exposed. Data involved: personal data and financial. Many organizations were affected.
Severity is rated High, calculated from how far the incident reached and how sensitive the data was.
Was the AI at fault?
Unclear. Sources do not establish whether the model's own behaviour or the surrounding systems were to blame.
Status and timeline
Confirmed by the affected organization, the AI vendor, a regulator or a named security research firm.
It became public on 30 Sep 2026.
The Financial Services Commission ordered firms to inspect every internet-facing system and close gaps by 8 October and to stop leaving personal credit data viewable by outside staff. It said no passwords or one-time codes were taken and warned customers about follow-up loan scams. Police opened a formal investigation.
Sources
- https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/
- https://therecord.media/south-korean-bank-hacks-ai-agents
- https://www.americanbanker.com/news/ai-linked-hacks-hit-korean-banks-through-loan-agent-sites
- https://www.koreatimes.co.kr/business/banking-finance/20261007/explainer-how-ai-emerged-as-new-threat-in-koreas-bank-hacking-crisis
- https://www.theregister.com/cyber-crime/2026/10/08/crowdstrike-finds-possible-bank-hackers-cv-among-exposed-ai-logs/5301908