AI Security
AI Pentest Tools and Open-Weight Models Used in Attacks: The Disclosed Facts and How to Prepare
In October 2026, CrowdStrike reported attacks that combined ARTEX, an open-source AI penetration-testing tool developed in China, with models such as DeepSeek v4.1-flash. Together with NIST's and Anthropic's assessments of GLM-5.3, this article separates the disclosed facts from what is still unknown, and sets out six things companies should prepare.
2026-10-20Ryuta Hamamoto