Chinese Hackers Use AI-Powered ARTEX to Target Over 200 Organizations Worldwide
Chinese state-sponsored hackers have used an AI-powered penetration testing framework called ARTEX to target over 200 organizations worldwide since October 17, 2026.
The activity has been attributed by CrowdStrike to a China-based threat actor tracked as Vapor Panda, which overlaps with a cluster detailed by Trend Micro as Earth Estries. First identified in February 2021, the group has leveraged the AI tool to conduct simultaneous attacks against high-value targets globally.
The campaign has specifically targeted foreign ministries in Latin America, Europe, Asia, and the Middle East; semiconductor, defense, and think tank organizations in the U.S.; and telecommunications providers in multiple regions. Post-exploitation activity has been confirmed at nine organizations, allowing the attackers to deploy BLOODALCHEMY, a derivative of ShadowPad.
AI-Powered Operations
Central to the operation is ARTEX, which combines DeepSeek and a commercial large language model (LLM) to conduct intrusions at scale. The framework uses planner and worker agents that coordinate with each other to move from initial access to reconnaissance and lateral movement, with a human operator overseeing the process.
Agentic tooling like ARTEX enables this scale of expansion at speed, allowing a small team to pursue a wider target set than manual operations would support, rather than a wholesale shift away from the adversary's established access-development mission.
The AI framework has enabled the attackers to pursue multiple attack paths in parallel, spanning web applications, cloud services, container environments, Active Directory, and mail infrastructure. When one path fails, they shift to alternatives, significantly expanding their capabilities compared to manual operations.
Sophisticated Techniques
The attacks have involved chaining security flaws, such as an unsigned SAML assertion and an XML external entity (XXE) vulnerability, to query cloud instance metadata services. This allowed the attackers to obtain roles that could issue administration commands to thousands of instances before escalating access.
Other techniques have included deploying web shells, abusing unauthenticated orchestration services to gain root access in container environments, reading Local Administrator Password Solution (LAPS) passwords, and leveraging resource-based constrained delegation to impersonate privileged accounts.
In one case, the attackers used authentication keys that hadn't been rotated since 2022 to forge an administrator token and access mailboxes. They then recovered VPN configurations from a user's sent mail and used VPN client keys from a 2024 email leak to reach internal networks.
Growing Threat Landscape
The disclosure coincides with the emergence of four previously undocumented China-linked state-sponsored threat groups: Hollow Panda, Overcast Panda, Titan Panda, and Veiled Panda. These groups have targeted various sectors worldwide for intelligence gathering and intellectual property theft.
Overcast Panda has been linked to close-access intrusions where they deployed their proprietary FlowCloud backdoor via direct physical access when targets were in China. Meanwhile, Titan Panda has targeted legal billing software by exploiting a hard-coded IIS machine key vulnerability in Intapp Time Server.
As ARTEX gains broader adoption among threat actors, its developer has taken it closed source. The framework's ability to orchestrate multiple LLM backends to autonomously perform reconnaissance and exploitation has significantly lowered the technical barrier for conducting complex intrusion campaigns.
Related
