South Korean Banks: 25,000 Customer Records Stolen in Attacks Powered by Open-Source AI Pentesting Tool ARTEX

A campaign against South Korean financial firms used ARTEX, an open-source agentic AI penetration-testing tool, together with large language models to steal customer data from institutions including Shinhan Bank and Yegaram Savings Bank, according to CrowdStrike Intelligence findings reported by The Hacker News on Oct. 8, 2026. The activity ran from late September to early October 2026 and resulted in data exfiltration from South Korea-based financial organizations. The breaches prompted an emergency response from South Korea's financial regulators, who confirmed leaks affecting tens of thousands of customers and warned consumers to watch for phishing attacks and loan scams stemming from the leaked data.

What was the specific vulnerability?

The specific weakness was not a flaw in the banks' software but the abuse of a publicly available AI attack tool: ARTEX, an open-source agentic penetration-testing framework developed in China, was operated against the banks with commercial LLMs as its reasoning backend. According to The Hacker News, the attack infrastructure used a two-server architecture. A Hong Kong-based IP address served as the campaign backbone, while a second server at IP 38.244.50[.]120 hosted the ARTEX instance suspected of the attacks on the Korean firms. That instance ran DeepSeek v4.1-flash as its main LLM backend, supplemented by Z.ai's GLM-5.3 and SpaceXAI's Grok 4.6. CrowdStrike assesses that DeepSeek access was likely obtained through the LLM API reseller "xcai[.]pro." The campaign has not been attributed to any known threat actor, but the evidence points to a suspected Chinese-speaking operator motivated by financial gain. The disclosure fits a wider pattern: in a related disclosure, CrowdStrike reported on SCARLET LOOP, an LLM agent equipped with 46 automation tools that used GPT-5.6 for dork generation and GPT-5.5 for target classification, showing that agentic AI tooling is now a recurring component of offensive operations.

How was the incident detected?

CrowdStrike Intelligence detected the campaign after finding open directories hosted at the Hong Kong IP address that exposed Claude Code session histories, Claude memory files, and ARTEX configuration files. Those exposed artifacts gave investigators a direct view into the operator's workflow. In the Claude Code sessions, the threat actor asked Claude where threat actors typically sell Korean data breach information and asked for help finding Korean Telegram data sales groups. One session referenced the Telegram account "@YY520CN" and the name "YY"; the same username appears in separate sessions on vulnerability research of a Telegram-based NFT gift marketplace. CrowdStrike cautions that these details cannot be definitively tied to the actor. Separately, South Korea's Financial Services Commission (FSC) held an emergency meeting after incident reports, and President Lee ordered a thorough investigation into personal data leaks at financial and public institutions, as reported by local media on Oct. 4 and 5, 2026, per BleepingComputer.

What is the current impact?

The confirmed impact is the theft of customer personal data and credit card information from at least four South Korean financial institutions, with regulators warning that the leaked data is expected to fuel phishing attacks and loan scams. Per BleepingComputer, citing local media and Yonhap, the FSC confirmed breaches at Shinhan Bank, where details of 25,000 customers were leaked, and at Kookmin Bank, where the credit card information of 119 clients was exposed, along with a limited breach at Hana Bank affecting a sales-support system. Each of Shinhan Bank and KB Kookmin holds more than $400 billion in assets. Mun Chong-hyun, head of the Genians Security Center, posted on LinkedIn that several threat analysts believe the breaches involved AI-based attack automation tools. On Oct. 8, 2026, the Telegram account @YY520CN denied involvement, stating: "I didn't do anything. South Korea just wants to blame all the untraceable things on me and other innocent ordinary people..."

What are the next steps?

The next steps are on-site investigations by South Korean authorities and mandatory security reviews across the financial sector, alongside the removal of the tool itself from open availability. South Korean financial companies have been instructed to inspect all externally accessible IT systems, reduce information exposure, fix missing or inadequate authentication and access controls, share threat information, and submit the results of internal security inspections. The ARTEX developer, Autumn-27, responded to the abuse by taking the project closed source, writing: "ARTEX was originally designed for the purpose of learning and research... In view of the reality of tool abuse, the ARTEX project will no longer be updated and will be converted to a closed source. There will be no release of any version or maintenance support in the future." The FSC and Financial Supervisory Service have urged consumers to watch for phishing attacks and loan scams stemming from the leaked data. No attribution has been confirmed.

By the numbers

Source snapshot