Cybersecurity
Over 800 exposed Artex AI hacking tools left vulnerable on the internet
Hundreds of instances of the Chinese-made autonomous penetration testing platform were discovered accessible without authentication, raising concerns over misconfigured cybersecurity infrastructure.
TechCrunch — AI
Security researchers have uncovered more than 800 publicly accessible instances of Artex, a Chinese-origin AI-powered penetration testing platform that can autonomously probe systems for vulnerabilities and conduct coordinated cyberattacks with minimal human oversight.
What is Artex?#
Artex is an open-source autonomous penetration testing tool—software designed to systematically identify weaknesses in computer networks and systems. Because it was released as free, open-source software in July, it has become widely accessible to attackers who lack direct access to proprietary cybersecurity AI models from providers like Anthropic and OpenAI. According to Microsoft's security researchers, Artex functions as a system of AI agents capable of orchestrating attacks from initial reconnaissance through data theft, often requiring little to no hands-on intervention.
The discovery and scope of exposure#
Cybersecurity firm Censys, which continuously scans the internet for exposed infrastructure, published findings showing that more than 800 distinct Artex deployments were accessible directly from the web without requiring passwords or login credentials. Silas Cutler, principal security researcher at Censys, noted that while some of these instances support legitimate authorized security testing, the sheer number of unprotected deployments raises questions about why they were ever exposed to the internet in the first place—the tool does not inherently require internet-facing access.
The exposed systems contained not only the tool itself but also attack infrastructure and malicious payloads, including implants designed to establish persistent footholds within compromised networks.
Known attacks leveraging Artex#
Security firms have linked Artex deployments to multiple large-scale operations. In September, attackers targeted South Korea's financial sector; CrowdStrike attributed those intrusions to a Chinese-speaking threat actor who had inadvertently left the Artex system in an "open directory." Censys' data revealed that the same exposed instance had been publicly available for months and contained attack records pointing to targets far beyond Korean financial institutions—including payment processors and gaming firms.
A second significant breach occurred in October when attackers exploited Artex to compromise ConnectWise, an IT services platform. That intrusion provided Chinese-linked adversaries with access to cloud-hosted environments across thousands of organizations using Microsoft's infrastructure, allowing them to exfiltrate sensitive data at speeds that left defenders insufficient time to mount an effective response.
