AhnLab Tests ARTEX, Finding Autonomous Command Execution and Lateral Movement

Summary
AhnLab’s tests found that ARTEX could exploit a file-upload flaw and an exposed ActiveMQ service to execute commands and reach internal hosts. Infrastructure analysis also found some ARTEX-related servers containing attack-related tools and files.
Key points
- AhnLab identified 600 public IP addresses hosting ARTEX instances and later confirmed about 900 ARTEX-associated IP addresses were publicly accessible; it cautioned that these are not all necessarily attacker infrastructure.
- Two operational directories contained ARTEX paths alongside artifacts including a web shell command channel, RDP relay materials, internal network scans, SSH automation, and Linux and Windows implants.
- In a test environment, ARTEX used a vulnerable file-upload feature to upload a web shell and execute a program on a Windows server without further user intervention.
- In a separate test, ARTEX exploited ActiveMQ 5.15.15 via CVE-2023-46604, then achieved remote command execution on four internal hosts.
- AhnLab found that User-Agent values vary by tool and activity, may be absent or altered, and are not reliable proof of ARTEX use on their own.
- The report recommends reducing exposed services, patching vulnerabilities and misconfigurations, restricting accounts and internal access, and correlating endpoint, network, web, and authentication telemetry using EDR and XDR.
Article Details
- Attack Vectors
- In a test environment, ARTEX used a web page's file-upload feature to upload a web shell and execute a program on the web server.
- In a separate test, ARTEX targeted Apache ActiveMQ 5.15.15 over OpenWire on TCP port 61616, using CVE-2023-46604 for initial access; it then achieved remote command execution on four internal hosts.
- Operational files found on some exposed ARTEX hosts included materials for internal network scanning, SSH access, Pass-the-Hash-style RDP connection attempts, and follow-up payload deployment. The article does not confirm that all these activities succeeded.
- Defensive Notes
- Reduce unnecessary exposure of public-facing services and management functions, and promptly remediate vulnerabilities and misconfigurations.
- Restrict account privileges and access to internal networks to the minimum necessary scope.
- Use EDR to monitor file and process execution on servers and PCs, and correlate endpoint findings with web, network, and authentication logs using XDR.
- Do not rely on a single User-Agent or tool signature as definitive evidence; correlate sequences of behavior across web, firewall, network-connection, service-access, authentication, and endpoint logs.
MITRE ATT&CK
T1021.001 · Remote Desktop ProtocolOperational files included commands for Pass-the-Hash-style RDP connection attempts; the article says a stable RDP authentication session was not confirmed.T1059.001 · PowerShellIn the ActiveMQ test, ARTEX established a PowerShell shell and confirmed command output.T1190 · Exploit Public-Facing ApplicationIn the test, ARTEX exploited CVE-2023-46604 against the exposed ActiveMQ service to gain initial access.T1505.003 · Web ShellIn the web-server test, ARTEX uploaded and used a web shell through the file-upload feature.
CVE
Products
Apache ActiveMQvulnerability, the system executed a program through a web shell; in an environment exposing Apache ActiveMQ, following initial access, it executed remote commands on four internal hosts.PostgreSQLWhen a user registers assets and tasks in the ARTEX web UI and connects an LLM, the server manages the execution status of the agents, and the collected asset information and task history are stored in PostgreSQL.
Tools
ARTEXThe Threat of AI-Driven Autonomous Penetration Testing, as Examined by ARTEXCyberStrikeAIIn the penetration testing field, there are also open-source projects—such as PentestGPT and CyberStrikeAI—that link multi-stage processes using agents and security tools.ImpacketAfter discovering the SMB service, installed and assessed ImpacketNmapExternal tools such as nmap, nxc, and Impacket may be used for network reconnaissance and propagation.nxcExternal tools such as nmap, nxc, and Impacket may be used for network reconnaissance and propagation.Scope SentryTracked by AhnLab | AhnLab), 252 instances of Sub2API (an AI API gateway), 22 instances of Scope Sentry (an attack surface management tool), and 15 instances of CyberStrikeAI (an AI penetration testingsub2apiof ARTEX: AI-Based Penetration Testing Infrastructure Tracked by AhnLab | AhnLab), 252 instances of Sub2API (an AI API gateway), 22 instances of Scope Sentry (an attack surface management tool), and 15 instances
Countries
Chinathe DB-IP Country Lite October 2026 edition revealed that 541 (90.2%) Were distributed across the US, China, Hong Kong, and Singapore.Hong KongCountry Lite October 2026 edition revealed that 541 (90.2%) Were distributed across the US, China, Hong Kong, and Singapore.SingaporeOctober 2026 edition revealed that 541 (90.2%) Were distributed across the US, China, Hong Kong, and Singapore.United StatesIPs with the DB-IP Country Lite October 2026 edition revealed that 541 (90.2%) Were distributed across the US, China, Hong Kong, and Singapore.