AI Red-Teaming Tools Are Now Being Used by Real Attackers

Security researchers have documented attackers actively using AI-powered red-teaming and penetration testing tools like ARTEX and CyberStrikeAI in real-world intrusions. These tools, originally developed for legitimate security testing, are now being deployed by malicious actors against organizations in retail, consulting, and other sectors worldwide.

The shift is significant because these tools are often free, easy to deploy, and require little more than access to a large language model. They allow attackers to autonomously evaluate targets, adapt when defenses are encountered, and persistently pursue objectives—particularly exploiting identity workflows and APIs exposed to the internet. The author of ARTEX has already taken the repository private to curb misuse, but new alternatives are emerging just as quickly, signaling a new phase in automated cyber threats.