Security researchers have uncovered hundreds of copies of a Chinese-developed autonomous AI hacking tool exposed directly to the web without requiring any password authentication. This digital exposure creates severe risks, potentially allowing malicious outsiders to launch cyberattacks against unsuspecting organizations and individuals across the globe.
The Capabilities of Artex and Microsoft’s Analysis
The software, known as Artex, is an open-source autonomous penetration testing tool designed to automatically probe computer systems for vulnerabilities. Since its public release in July, it has been linked to multiple major cyberattacks. Because it is open-source, anyone can download and run it on their own servers, making it a popular option for individuals who lack access to advanced cybersecurity AI models from providers like Anthropic or OpenAI. Microsoft previously described Artex as a system of AI agents capable of executing cyberattacks spanning from initial reconnaissance to data exfiltration with minimal human intervention.
Censys Discovery of Unprotected Deployments
In a published blog post, the internet-scanning cybersecurity firm Censys revealed that it identified more than 800 distinct deployments of Artex accessible from the web without a login password. Silas Cutler, the principal security researcher at Censys who authored the post, noted that not every instance is operated maliciously, as many serve authorized testers. However, the exact reason why hundreds of these systems are exposed remains unclear, as the software does not inherently require internet accessibility to function.
Misconfigurations and Large-Scale Cyberattacks
These findings illustrate how powerful attack tools can be dangerously misconfigured, increasing the likelihood of unauthorized discovery and misuse. The exposed systems also leak offensive tools such as hidden implants designed to compromise and maintain persistent access to victim networks. Security firms have tied Artex deployments to several broad-scale campaigns, including attacks against the South Korean financial sector in September, which CrowdStrike attributed to a Chinese-speaking hacker.
ConnectWise Breach and Ongoing Threats
Censys noted that the specific Artex instance linked to the South Korean attacks had been publicly exposed for months and contained data extending far beyond banking institutions into gaming and payment sectors. Furthermore, Artex was utilized in an attack on IT support provider ConnectWise in October, allowing Chinese-linked hackers to infiltrate Microsoft-hosted cloud environments and exfiltrate data so rapidly that defenders had virtually no time to respond. Many related attacks exploited basic weaknesses such as weak credentials and the absence of multi-factor authentication.
In response to these security concerns, Artex announced a transition to a closed-source model to limit distribution to paid customers. Nevertheless, numerous copies of the code remain available on code-sharing platforms, ensuring that the software can still be deployed in active malicious campaigns.
