Recent investigations by cybersecurity experts have revealed that numerous instances of a Chinese-developed AI hacking tool have been left exposed online, posing risks as they may facilitate cyberattacks against unwitting victims.
The software in question, named Artex, is an open-source autonomous penetration testing tool designed to automatically identify vulnerabilities in computer systems. Since its launch in July, it has been associated with various cyber incidents. Being open-source allows anyone to download and utilize it freely, rendering it an appealing alternative for users who lack access to more sophisticated cybersecurity AI models from companies like Anthropic or OpenAI. Microsoft has characterized Artex as a system of AI agents capable of executing cyberattacks with minimal human oversight, covering aspects from reconnaissance to data extraction.
Censys, a cybersecurity firm that monitors the internet for exposed systems, reported finding over 800 separate instances of Artex that were reachable online without any login requirements.
According to Silas Cutler, a lead security researcher at Censys and author of the report, not all instances of Artex are necessarily utilized for malicious purposes; many are employed by authorized testers. However, the reason behind the public exposure of so many Artex systems remains unclear, especially since the software can function without being accessible on the internet.
These findings highlight the risks posed by powerful hacking tools that can be wrongly configured for public access, increasing the likelihood of misuse. Furthermore, these publicly accessible systems may disclose cyberattack tools, including implants designed to compromise networks and maintain ongoing unauthorized access.
Artex has been linked to several significant cyberattacks, such as one targeting the South Korean finance sector in September. Cybersecurity firm CrowdStrike attributed this attack to a Chinese-speaking hacker who exploited an Artex instance, which was poorly secured to have an “open directory.” Censys indicated that the exposed Artex instance had been accessible for months, containing data relevant to attacks that extended beyond just Korean banks, affecting payment and gaming sectors as well.
Additionally, Artex was implicated in a recent attack on IT support company ConnectWise in October, enabling hackers connected to China to surveil thousands of organizations utilizing Microsoft’s cloud services. Microsoft, with insight into the attack, noted that the hackers swiftly breached customer cloud setups and exfiltrated data before defenders could effectively intervene.
Numerous attacks associated with Artex exploited straightforward vulnerabilities, including weak password practices and the absence of multi-factor authentication, which further emphasizes that basic security oversights can endanger both attackers and their targets.
In response to these issues, Artex has announced it will transition to a closed-source model, limiting further distribution to paying clients. Nonetheless, there remain several copies of Artex’s code available on code-sharing repositories, which could still facilitate active cyberattacks.
