Open Doors: Over 800 Copies of Autonomous AI Hacking Tool Found Left Wide Open Online

Security researchers discovered hundreds of publicly exposed installations of Artex, a Chinese-made artificial intelligence hacking tool, sitting online without basic password protection. The open availability allows outside actors to access live attack consoles and launch cyberattacks against targets around the globe.

Artex is an open-source penetration testing framework built to automate security checks by scanning systems for vulnerabilities. Since its public release in July, bad actors have connected the tool to several high-profile cyberattacks. Because Artex was released as free software for anyone to download and run on private servers, it quickly became a popular choice for hackers who lack access to top-tier commercial security models. Software engineers at Microsoft described Artex as a network of automated software agents that run full cyberattacks, carrying out everything from early system reconnaissance to data theft with little human oversight.

Cybersecurity firm Censys uncovered the leak after scanning the public web for exposed interfaces. Researchers identified more than 800 distinct deployments of Artex accessible directly through web browsers without requiring any user credentials. Silas Cutler, principal security researcher at Censys, explained that while security teams use some exposed instances for legitimate internal testing, leaving these setups open to the public internet creates massive risk. The underlying software does not require internet-facing web access to run effectively, making the public exposures entirely unnecessary.

Leaving these consoles exposed highlights how poor server configurations turn internal tools into active security risks. Unprotected setups give unauthorized users a chance to hijack active attack frameworks, exposing hidden malicious implants, custom exploit code, and persistent access backdoors built into target networks.

Security firms previously tied Artex to major attack campaigns across multiple industries. In September, hackers targeted the South Korean financial sector using the tool. Analysts at CrowdStrike traced those intrusions back to a Chinese-speaking hacker who left an open directory hosting the Artex software. Investigation data showed that the exposed instance stayed online for months, holding stolen records that extended far beyond South Korean banks to reach international payment processors and online gaming companies.

Hackers also deployed Artex during an October attack against IT management firm ConnectWise, using autonomous agents to spy on thousands of client organizations. Microsoft security teams tracking the incident noted that attackers used Artex to breach customer cloud environments and steal stored files so fast that human defenders had little time to catch the intrusion or stop data theft.

Many intrusions linked to Artex did not rely on complex exploits. Instead, the automated agents targeted simple vulnerabilities like weak passwords, default system logins, and unconfigured multi-factor authentication controls. The software accelerates basic reconnaissance tasks, letting bad actors locate exposed systems in a fraction of the time required by manual scanning methods.

Following public reports, the creator of Artex pulled the original code repository and announced plans to turn the project into closed-source software. However, dozens of archived copies and rebranded versions still circulate across public code-sharing platforms. Security teams warn that lingering copies will remain active in cyberattack campaigns as long as operators leave server ports open and default configurations unpatched.