Cybersecurity Research

AI-Powered Red Teaming Tools Like ARTEX Are Being Turned Against Real Targets

Unit 42 · 12 Nov 2026
Key Takeaway Assume attackers can now probe your internet-facing logins and APIs automatically and at scale, so protect those entry points and the cloud accounts behind them with strong authentication and close monitoring.

Palo Alto Networks' Unit 42 says a fast-growing group of AI-enabled tools, built and marketed for red teaming and penetration testing, is now being used by attackers to carry out offensive operations on their own. Two examples are ARTEX and CyberStrikeAI. ARTEX is an open-source, multi-agent AI framework first published in mid-2026 and widely copied since. It is free, quick to set up and needs little more than access to a large language model.

Unit 42 says its Incident Response team investigated multiple intrusions where the attacker's server hosted an ARTEX console or used infrastructure and user agents linked to the tool. Its Managed Threat Hunting team also tracked a coordinated ARTEX campaign against customer cloud environments worldwide, including in retail and consulting. As of October 2026, Unit 42 identified 1,323 internet-facing hosts running CyberStrikeAI and 609 running ARTEX, with 21 running both. The tools can adapt when something fails and keep pursuing their goal, and can target identity workflows and internet-exposed APIs, then use stolen credentials and trust relationships to reach the cloud.

The ARTEX author has since closed the source and stopped public development to limit misuse, but Unit 42 notes that alternatives are filling the gap. It analysed 14 tools in total. Its central warning is that attackers no longer need specialised skills to run adaptive, multi-stage intrusions, and that financially motivated actors can manage dozens of concurrent attacks at minimal cost using models such as DeepSeek.

Summarised by CISO AI from Unit 42, written with Claude Sonnet 5.5. We link back to every original so you can read it yourself.