Artex becomes criminals' new favorite "foolproof" AI hacking tool
Artex Is Criminals’ New Favorite ‘Point-and-Shoot’ AI Hacking Tool
Several security vendors and the FBI say that since around late August, criminals have used AI penetration frameworks to attack multiple companies, government agencies and banks worldwide, with the open-source framework Artex the most prominent. Artex was released on GitHub in July by a developer using the pseudonym Autumn-27; it is not an AI model but a framework that connects to external frontier and open-source models via API and calls existing penetration testing tools, letting users direct multiple agents through a full attack workflow with natural-language prompts. CrowdStrike says the China-linked hacking group Vapor Panda used it in October to attack more than 200 organizations; Microsoft found that Storm-0940 used it after a supply chain attack to attempt logins to Microsoft 365 and gained privileged access in under three hours. The developer took it down on October 8, saying it was for educational research only, but copies are still circulating. Researchers note that such tools are not currently stealthy, and basic security measures such as patching and not using default credentials can stop most attack techniques.
AI penetration frameworks such as Artex have been used in real attacks and have an extremely low barrier to entry; security vendors point out that such tools are currently not stealthy and that basic defenses can block most techniques.