The FBI warns that cybercriminals and state-aligned hackers are using ARTEX, an open-source AI penetration-testing framework, to automate attacks against organizations worldwide. Some intrusions have progressed from initial targeting to privileged access (elevated permissions) in under three hours, according to the bureau’s FLASH advisory. That’s an unpleasantly short response window. SourceSource
The advisory, listed by IC3 under November 13, 2026, was amplified by FBI Dallas on November 17. The bureau says malicious actors have used ARTEX since at least August, targeting financial services, government, defense, energy, healthcare and other sectors. SourceSourceSource
How ARTEX coordinates attacks
According to the advisory, ARTEX translates natural-language objectives into commands executed by specialized agents, requiring minimal human involvement after the initial objective is set. It connects to external language models rather than containing its own model. The developer said it was intended for authorized security testing, the FBI reports; attackers have repurposed it to automate work from reconnaissance through data theft. Source
The bureau says the framework lets operators with limited technical expertise run complex operations against multiple targets simultaneously. It can generate noisy traffic that triggers alerts, but its speed can leave defenders little time to respond. Loud, unfortunately, doesn’t mean slow. Source
FBI Dallas points to the October ConnectWise supply-chain compromise, saying it exposed Microsoft 365 customer environments and involved attempted authentication against more than 2,250 downstream tenants. That figure describes tenants facing authentication attempts—not confirmation that every one was successfully breached. Source
The advisory says the developer took the original GitHub repository offline on October 8 following reports of abuse, but community forks, archived copies, localized variants and a commercial version remained available. Repository removal was not an off switch. Source
Stolen keys and deepfake attempts
The Register reports that Amazon Threat Intelligence observed two ARTEX campaigns between October 3 and November 5. In one, attackers stole AI access keys and fed them into the framework to power more agents searching for more keys: a particularly grubby feedback loop. The reporting also describes attackers abusing cloud credentials to access models. Source
Comments