AI Hacking Frameworks Exposed: 1,428 Services Found Online

• •1,428 services from the ARTEX family were found online, indicating widespread exposure.
• •Many AI hacking frameworks lack authentication on their management consoles, increasing vulnerability.
• •The proliferation of forks from the ARTEX project continues despite its takedown.

Researchers have identified 45 AI hacking frameworks, including ARTEX, PentAGI, and Agent Zero, running on internet-accessible servers. A Shodan scan on November 7, 2026, revealed 1,428 services associated with the ARTEX family, 419 for PentAGI, and 173 for Agent Zero. These frameworks connect large language models to offensive security tools, enabling potential exploitation. Notably, many of these frameworks have management consoles that lack authentication, increasing the risk of unauthorized access. The takedown of ARTEX did not halt the proliferation of its forks, with 102 repositories created in the week following its shutdown. The frameworks are capable of performing reconnaissance and post-exploitation tasks, raising concerns about their use in cyberattacks. The situation remains critical as the spread of these tools continues to pose significant risks to organizations.