An actor used the agentic pentest tool ARTEX and Claude Code to breach South Korean financial organizations, CrowdStrike reports
From late September to early October 2026, an actor CrowdStrike assesses as a likely financially motivated Chinese speaker used the open-source agentic pentesting tool ARTEX with DeepSeek v4.1-flash, plus Claude Code sessions, to breach South Korean financial organizations and steal data.
From late September to early October 2026, a targeted campaign against South Korean financial organizations "resulted in exfiltrated data." At one bank the actor breached a loan progress inquiry service used by financial brokers; at another it compromised an employee mobile work-support system. The number of affected organizations is unconfirmed1.
The actor ran ARTEX, "a recently released open-source agentic penetration testing (pentesting) tool developed in China," with DeepSeek v4.1-flash as its main LLM, and used GLM-5.3 (Zhipu AI) and Grok 4.6 for additional Claude Code sessions. Claude Code session histories, ARTEX configuration files, and Claude memory files turned up in open directories the actor controlled1.
CrowdStrike, which reported the campaign on October 7, 2026, hasn't named the adversary but assesses with moderate confidence that it's a financially motivated Chinese speaker1.
CrowdStrike reported a targeted campaign against South Korean financial organizations that resulted in exfiltrated data.
“a targeted campaign against South Korean financial organizations that resulted in exfiltrated data”
According to CrowdStrike, the campaign was active from late September to early October 2026.
“The campaign was active from late September to early October 2026.”
CrowdStrike says that at one bank the actor reportedly breached a loan progress inquiry service used by financial brokers.
“At one affected bank, the threat actor reportedly breached a loan progress inquiry service used by financial brokers.”
CrowdStrike says that at another bank the actor compromised an employee mobile work-support system.
“At another bank, the threat actor compromised an employee mobile work–support system.”
CrowdStrike says the number of affected organizations remains unconfirmed.
“As of this writing, the number of organizations affected remains unconfirmed.”
CrowdStrike says the actor used ARTEX, an open-source agentic penetration testing tool developed in China.
“ARTEX, a recently released open-source agentic penetration testing (pentesting) tool developed in China”
According to CrowdStrike, the actor's ARTEX instance used DeepSeek v4.1-flash as its primary LLM backend.
“The ARTEX instance used DeepSeek v4.1-flash as the primary LLM backend”
CrowdStrike says the actor also used GLM-5.3 (Zhipu AI) and Grok 4.6 for additional Claude Code sessions.
“the threat actor supplemented this LLM with GLM-5.3 (Zhipu AI) and Grok 4.6 for additional Claude Code sessions”
CrowdStrike found Claude Code session histories, ARTEX configuration files, and Claude memory files in open directories controlled by the actor.
“Analysis of threat actor-controlled open directories uncovered Claude Code session histories”
CrowdStrike has not attributed the activity to a named adversary. It assesses with moderate confidence that the actor is likely a Chinese speaker and financially motivated.
“the threat actor is likely a Chinese speaker and financially motivated”
All 10 claims with their quotes
- CrowdStrike reported a targeted campaign against South Korean financial organizations that resulted in exfiltrated data.1
“a targeted campaign against South Korean financial organizations that resulted in exfiltrated data”
- According to CrowdStrike, the campaign was active from late September to early October 2026.1
“The campaign was active from late September to early October 2026.”
- CrowdStrike says that at one bank the actor reportedly breached a loan progress inquiry service used by financial brokers.1
“At one affected bank, the threat actor reportedly breached a loan progress inquiry service used by financial brokers.”
- CrowdStrike says that at another bank the actor compromised an employee mobile work-support system.1
“At another bank, the threat actor compromised an employee mobile work–support system.”
- CrowdStrike says the number of affected organizations remains unconfirmed.1
“As of this writing, the number of organizations affected remains unconfirmed.”
- CrowdStrike says the actor used ARTEX, an open-source agentic penetration testing tool developed in China.1
“ARTEX, a recently released open-source agentic penetration testing (pentesting) tool developed in China”
- According to CrowdStrike, the actor's ARTEX instance used DeepSeek v4.1-flash as its primary LLM backend.1
“The ARTEX instance used DeepSeek v4.1-flash as the primary LLM backend”
- CrowdStrike says the actor also used GLM-5.3 (Zhipu AI) and Grok 4.6 for additional Claude Code sessions.1
“the threat actor supplemented this LLM with GLM-5.3 (Zhipu AI) and Grok 4.6 for additional Claude Code sessions”
- CrowdStrike found Claude Code session histories, ARTEX configuration files, and Claude memory files in open directories controlled by the actor.1
“Analysis of threat actor-controlled open directories uncovered Claude Code session histories”
- CrowdStrike has not attributed the activity to a named adversary. It assesses with moderate confidence that the actor is likely a Chinese speaker and financially motivated.1
“the threat actor is likely a Chinese speaker and financially motivated”
Sources
Editor notes
Single-vendor account (CrowdStrike); no victim, regulator, or news confirmation sourced yet. CrowdStrike does not say how much of the intrusion work the AI tooling did on its own versus the human operator, so how central the AI was is inferred from the use of an agentic pentesting framework and should be checked by a reviewer. GLM-5.3 and Grok 4.6 ran in Claude Code sessions per CrowdStrike; Claude Code was used as the agent harness with non-Anthropic models, so Anthropic is not listed as developer. The page notes the bank breach "reportedly"; types of data taken are not stated. Look for Korean news coverage (e.g. bank breach disclosures from late Sept/early Oct 2026) to corroborate.