FBI Links ARTEX Framework to Multisector Cyber Intrusions
Malicious actors have used the ARTEX penetration-testing framework in cyber campaigns across multiple countries since at least August 2026, according to an FBI FLASH alert published on November 13, 2026. The bureau said some intrusions involving the tool reached privileged access less than three hours after initial targeting.
ARTEX uses multiple software agents to carry out testing tasks. According to the FBI, the open-source framework translates objectives written in ordinary language into commands for specialized agents, requiring little human involvement once an objective has been defined. Rather than including a language model of its own, ARTEX can connect to external models.
The FBI identified activity against organizations in finance, government and technology, as well as defense, energy, health care, manufacturing, telecommunications and legal services. As an example of earlier ARTEX-linked activity, the bureau cited a supply-chain compromise involving ConnectWise Cloud Backup customers. That compromise enabled access to more than 2,250 downstream tenants, according to the alert.
CERT-EU said ARTEX was designed to let operators define the boundaries of testing and guide its operation. However, it said compliance with those boundaries and avoidance of destructive actions rely on adjustable rules and the model, not technically enforced restrictions.
The framework remained available in alternative forms after its original GitHub repository became unavailable on October 8, 2026, following reports of misuse, the FBI said. Copies and independently developed versions were still accessible through repository forks and mirrors, according to the bureau.
By early November 2026, researchers had identified over 1,600 publicly reachable ARTEX consoles, according to the FBI. The alert said the number of deployments could exceed that count because consoles on private networks or behind firewalls would fall outside the researchers’ tally of internet-exposed systems.
For organizations reviewing the alert’s indicators, the FBI recommended investigation or verification before taking action. The bureau noted that IP addresses used by cloud-hosted or virtualized infrastructure can be reassigned. An address associated with earlier malicious activity may therefore identify a benign service at a later point, according to the alert.
* This article is based on publicly available information at the time of writing.