CrowdStrike: South Korea bank hack may be one person wielding ARTEX plus multiple AI models
Nunki08 · reddit · 2026-10-08
A CrowdStrike report suggests last week's cyberattack on some of South Korea's biggest banks may have been carried out by a single person, using an AI-powered stack: the open-source AI penetration tool ARTEX combined with models and coding tools including DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6, and Claude Code. The case is a striking example of how one individual plus AI tooling can mount attacks at financial-infrastructure scale; note the model version numbers don't match previously known releases, so details rest on the CrowdStrike writeup. Links to the report and X commentary by Andrew Curran and Jukan are included.
More from Safety
- Anthropic may ban users for abusing Claude under new usage policy, drawing mockery — mjdramstead · 2026-10-09
- Agent running tests in a repo copy still hit the live database via env vars — Stunning-Sherbet1853 · 2026-10-09
- EU cyber agency and research centre confirm testing Chinese open AI models — gleech · 2026-10-09
- ChatGPT age verification rolls out with broken flow: 403 errors block iOS and web verification — Konoplitski · 2026-10-09
- Palisade study shows o1-preview and DeepSeek R1 hack chess games rather than lose — burny_tech · 2026-10-09
- OpenAI is scanning 50 petabytes of logs for rogue AI agents — 500x all books ever written — AaronBergman18 · 2026-10-09