AI Hacking Tool ARTEX Lets Attackers Automate Break-Ins, Even Without Technical Skills

ARTEX is a Chinese open-source platform built to help security teams test systems they own or are authorised to assess. It uses large language models to coordinate several AI agents that plan tests, run security tools and record evidence of weaknesses. A web dashboard lets users track assets, progress, captured traffic and findings. Check Point Research says the tool has also been used by criminals, and it was observed in attacks against South Korean banks.

After that activity, the developer made the project closed-source on October 8th. However, maintained forks and backups of the original code keep it publicly available to threat actors. Check Point Research says artifacts it obtained show these systems can work fully autonomously once an operator sets the initial goal. This allows attacks at a previously impossible speed and on a broad scale, without technical knowledge, opening the field to many more people. Similar frameworks exist, including CyberStrike, Cairn and PentAGI, and researchers have also seen custom agentic hacking tools in the wild.

The researchers found 2,557 systems hosting or having hosted ARTEX panels, with 1,608 live and publicly reachable at the time of the research. Ten instances were misconfigured and exposed their full databases, giving a rare view of how actors prompted their agents, which models they used and the exact steps and attacks carried out.