Agentic Penetration Testing Framework Abused in Attacks on Critical Infrastructure Entities
Cybercriminal actors are using a weaponized version of the ARTEX autonomous penetration testing framework to conduct attacks on organizations in multiple sectors, including healthcare. ARTEX is an open-source, large language model (LLM)-driven agentic penetration testing platform that was developed for red team assessments to identify security weaknesses.
The tool was released on GitHub by developer Autumn-27 in July 2026, and it was rapidly adopted. The first known instances of the tool being weaponized for use in offensive cyber operations occurred in August 2026. Following reports of abuse by threat actors, the project was shut down by the developer in early October; however, the platform remains widely accessible through independent routes and mirrors on code-sharing platforms. The Federal Bureau of Investigation (FBI) has observed ongoing threat actor adoption of the platform.
ARTEX is an autonomous, multi-agent penetration testing platform that translates high-level tactical goals described in natural language into actionable machine commands that are executed by artificial intelligence agents. While human involvement is required to set the initial objectives, further hands-on action is minimal. The platform can be used to automate multiple phases of enumeration and exploitation simultaneously, allowing threat actors to conduct attacks at a scale previously unavailable to them. ARTEX is particularly adept at identifying and exploiting vulnerabilities in peripheral systems that typically receive less security attention than flagship applications.
The FBI warns that attacks generate high volumes of forensically noisy traffic, and while security alerts may be generated, the speed at which systems can be attacked provides little time for a coordinated defender response. While ARTEX does not incorporate its own AI model, it integrates with external LLM providers including Claude, OpenAI GPT, and others, most commonly models of Chinese origin such as the DeepSeek family.
“ARTEX lowers technical barriers for conducting complex intrusions, enabling threat actors with minimal expertise to execute sophisticated campaigns,” explained the FBI in a recent FBI Flash advisory. The advisory shares observed Tactics, Techniques, and Procedures and Indicators of Compromise, along with recommended mitigations to improve security posture against ARTEX attacks and similar agentic AI frameworks.