FBI warns of growing abuse of AI-driven penetration testing framework ARTEX
FBI Highlights Growing Abuse of AI-Driven Pentesting Framework ARTEX
In a Flash advisory published on November 13, the FBI warned that multiple malicious actors have since at least August 2026 been using the open-source AI penetration testing framework ARTEX to attack organizations across finance, government, defense, semiconductors and other industries worldwide. Recent cases include an attack on South Korean financial institutions that stole the personal information of about 67,000 people, and a supply chain intrusion affecting ConnectWise Cloud Backup customers that gave attackers access to more than 2250 downstream tenants. ARTEX was released in July 2026 by GitHub user Autumn-27 and can connect to external LLMs such as Claude, GPT and DeepSeek; the FBI's analysis of attacker-controlled instances found Chinese models to be the most common, with the DeepSeek Flash series dominant because of its balance of capability and cost and its lack of observable guardrails in a penetration testing context. The FBI assesses that the tool can automate the enumeration and exploitation phases of an intrusion, that some intrusions went from initial target identification to privileged access in under three hours, and it provided mitigation recommendations.
The FBI advisory shows that some intrusions went from initial target identification to privileged access in under three hours; the tool's speed leaves defenders no time to coordinate a response.