Over 800 Unsecured Artex AI Hacking Tools Found Exposed Online
Researchers at Censys uncovered over 800 publicly exposed instances of Artex, an open-source AI penetration testing tool linked to major recent cyberattacks.
Security researchers at cybersecurity firm Censys have uncovered more than 800 active deployments of Artex, an autonomous AI-driven penetration testing tool, left completely open to the public internet. The exposed systems require no password to log in, leaving powerful automated hacking tools accessible to anyone on the web. Originally released as an open-source project in July, the Chinese-developed software allows users to run automated security assessments, but poor operational security by users has turned many of these instances into accessible attack vectors.
Unsecured AI Security Tools on the Open Web
Silas Cutler, principal security researcher at Censys who detailed the findings in a blog post, noted that while not all exposed instances are operated with malicious intent, there is no technical requirement for Artex deployments to be connected directly to the public web. Artex operates as a system of AI agents designed to automate cyberattacks, spanning initial reconnaissance, network penetration, and data exfiltration with minimal human involvement, according to previous analysis by Microsoft.
Because Artex was made freely available for anyone to self-host, it gained quick traction among individuals and teams who cannot access top-tier cybersecurity AI models developed by companies like Anthropic or OpenAI. However, the exposed instances do not just reveal the interface of the software; they also expose sensitive attack tools, including implants—hidden malicious software used to compromise networks and maintain persistent access to a victim's network.
Real-World Cyberattacks Linked to Misconfigured Systems
The public exposure of Artex tools has already been linked to major security breaches across global industries. In September, cybersecurity firm CrowdStrike tied an attack against South Korea’s financial sector to a Chinese-speaking threat actor who relied on an exposed Artex deployment. Censys later confirmed that the specific instance had remained accessible for months, containing data tied to intrusions that extended beyond banks to payment processors and gaming companies.
Furthermore, in early October, attackers utilized Artex during a breach of IT support giant ConnectWise. That intrusion allowed Chinese-linked hackers to spy on thousands of organizations whose files and data are hosted within Microsoft’s cloud infrastructure. Microsoft observed that the speed of the AI-driven attack allowed adversaries to breach customer cloud environments and steal data before defensive teams had time to detect or respond. Ironically, many of these high-tech AI attacks began by exploiting basic security oversights, such as simple credential guessing and missing multi-factor authentication.
What it means for developers
For software engineers and security teams, the Artex exposures serve as a stark reminder that advanced AI tools are subject to the same basic misconfiguration risks as conventional server software. Leaving management interfaces or command nodes unauthenticated on the public internet completely undermines the security posture of an organization, regardless of how sophisticated the underlying software is.
Developers seeking to integrate advanced AI capabilities into their security workflows or software projects must prioritize secure deployment architectures. While open-source tools provide accessibility, accessing top-tier frontier models securely via established infrastructure can mitigate the operational overhead of self-hosting sensitive toolsets. Developers can try top AI models cheaply through one API at https://apixoai.online, allowing teams to leverage advanced intelligence without risking exposed self-hosted nodes.
In response to the recent breaches and operational exposure, the creators of Artex announced plans to move the project to a closed-source model, restricting future releases to paid customers. Despite this shift, dozens of archived copies of the original open-source codebase remain widely available on code-sharing sites, meaning unsecured and active instances of the autonomous hacking tool will likely persist across the internet for the foreseeable future.
Source: Hundreds of Artex AI hacking tools found publicly exposed to the web — TechCrunch AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
