WSJ: hackers used open multi-agent pentest tool Artex to hack Japanese websites
Ok_Warning2146 · reddit · 2026-10-27
Per WSJ, hackers used Artex, a multi-agent penetration testing tool by China-based Autumn-27, to compromise multiple Japanese websites. The original GitHub repo has been deleted, but a Korean developer's fork with internationalization (artex-global) remains available.
The poster notes attackers likely paired the tool with open-weight models. Open weights are neutral in themselves, but in the hands of bad actors they can cause real damage — a concrete case of the open-source double-edged sword.
More from Safety
- Cloudflare: GLM 5.3 on Workers AI leaked other customers' data into 155 responses — djcroman · 2026-10-27
- Fake a '200 OK' and the agent just keeps going: an agent reliability experiment — ngxson · 2026-10-27
- Was Sydney misaligned? 2023 alignment analysis gains new relevance — jankulveit · 2026-10-27
- cve.now launches: a free CVE feed for AI agents with RSS and MCP support — Kyrannio · 2026-10-27
- Jailbreak prompt collection repo rockets past 4.8k stars, +890 in a day — togg53192-cmd · 2026-10-27
- AI's week in good and evil: 14-year-old PHP bug found, 2,250 firms attacked by AI-armed hackers — alex_verem · 2026-10-27