# Crowdstrike 报告疑似单人使用 AI 渗透工具攻击多家韩国银行
> 原标题:AI-powered hacking tools enabled a likely single attacker to breach multiple South Korean banks
- 来源:The Decoder:AI News
- 发布时间:2026-10-08T09:24:17.000Z
- AIHOT:https://aihot.news/items/br6vb46mhi9mkv2pxiq59k8z7
- 原文:https://the-decoder.com/ai-powered-hacking-tools-enabled-a-likely-single-attacker-to-breach-multiple-south-korean-banks/
## 摘要
Crowdstrike 报告称,一名疑似中文使用者于2026年9月底至10月初利用 AI 驱动的开源渗透测试工具 ARTEX 攻击多家韩国金融机构,窃取大量数据,其中 Shinhan Bank 超过 25,000 条包含姓名、联系方式、收入和信用额度的记录泄露。
## 推荐理由
原文给出了攻击工具、所用模型和泄露规模等具体细节,可帮助读者认识AI驱动的渗透测试如何放大单人攻击能力。
## 正文 · 中文译文
**Crowdstrike 报告了韩国的一起基础设施黑客攻击事件。** 一名疑似讲中文的攻击者在 2026 年 9 月下旬至 10 月初期间袭击了多家韩国金融机构,窃取了大量数据。据 [韩国报纸《韩民族》](https://www.khan.co.kr/en/article/202610042320007) 报道,仅新韩银行一家就有超过 25,000 条包含姓名、联系方式、收入和信用额度的记录被盗。韩国金融监管机构召开了紧急会议。总统李在明呼吁进行彻底调查。
攻击者使用了 [ARTEX](https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/),这是一款中国的开源工具,于 7 月首次发布在 GitHub 上,它利用 AI 语言模型进行自动化渗透测试,也就是说它能自行发现安全漏洞。其背后的模型是 DeepSeek v4.1-flash、GLM-5.3 和 Grok 4.6。研究人员在攻击者的开放目录中发现了 Claude Code 会话日志,显示其搜索 Telegram 群组以出售窃取的数据。
Crowdstrike 表示,此案表明 AI 工具如何能让一个人在短时间内实施大规模入侵,这正是 [网络安全专家数月来一直警告的风险](https://the-decoder.com/open-weight-models-now-match-frontier-cyber-performance-from-just-four-months-ago-at-a-fraction-of-the-cost/)。就在几天前,Anthropic 记录到 [GLM-5.3 编写漏洞利用程序的能力几乎与 Mythos Preview 相当](https://the-decoder.com/anthropic-says-zhipus-open-weight-glm-5-3-nearly-matches-claude-mythos-preview-at-building-exploits/),后者是 Anthropic 的前沿模型,也是 [在 2026 年 3 月下旬引发整个争论的那个模型](https://the-decoder.com/from-gpt-2-to-claude-mythos-the-return-of-ai-models-deemed-too-dangerous-to-release/)。
### 没有炒作的 AI 新闻——由人类精选
订阅 THE DECODER,享受无广告阅读、每周 AI 新闻通讯、每年六次我们独家的“AI Radar”前沿报告、完整档案访问权限以及评论区访问权限。
## 正文 · 原文
**Crowdstrike reports on an infrastructure hack in South Korea.** A suspected Chinese-speaking attacker hit multiple South Korean financial institutions between late September and early October 2026, stealing large amounts of data. At Shinhan Bank alone, more than 25,000 records with names, contact details, income, and credit limits were stolen, according to [Korean newspaper Khan](https://www.khan.co.kr/en/article/202610042320007). South Korea's financial regulator held an emergency meeting. President Lee Jae Myung called for a thorough investigation.
The attacker used [ARTEX](https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/), a Chinese open-source tool first posted on GitHub in July that uses AI language models for automated penetration testing, meaning it finds security flaws on its own. The models behind it were DeepSeek v4.1-flash, GLM-5.3, and Grok 4.6. Researchers found Claude Code session logs on the attacker's open directories, showing searches for Telegram groups to sell stolen data.
Crowdstrike says the case shows how AI tools can let a single person pull off massive breaches in a short window, the kind of [cybersecurity risk experts have been warning about for months](https://the-decoder.com/open-weight-models-now-match-frontier-cyber-performance-from-just-four-months-ago-at-a-fraction-of-the-cost/). Just days earlier, Anthropic documented that [GLM-5.3 can write exploits nearly on par with Mythos Preview](https://the-decoder.com/anthropic-says-zhipus-open-weight-glm-5-3-nearly-matches-claude-mythos-preview-at-building-exploits/), Anthropic's frontier model and [the one that sparked the entire debate in late March 2026](https://the-decoder.com/from-gpt-2-to-claude-mythos-the-return-of-ai-models-deemed-too-dangerous-to-release/).
### AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.