CrowdStrike says China-nexus group used AI agents against more than 200 organisations in 11 days

Security firm CrowdStrike said on 29 October that VAPOR PANDA, a China-nexus intrusion group, used an agentic AI framework called ARTEX in a campaign against more than 200 organisations in at least 14 countries. The activity began on 17 October and ran over an 11-day window. CrowdStrike says it is the first time it has seen the group use AI-driven intrusion tools.

CrowdStrike confirmed post-exploitation activity at nine organisations, where the attackers installed a backdoor, sprayed credentials and scanned networks. Password spraying and credential scanning reached at least 12 more. Targets included foreign ministries in several regions, US pharmaceutical, semiconductor, defence, nuclear and financial organisations, telecommunications providers and think tanks.

ARTEX runs a main agent with planner and worker agents that share a graph of facts and goals. According to CrowdStrike, it ran mainly on the Chinese DeepSeek model, with a second commercial model it could not identify, and in one intrusion the planning agents used about 1.4 billion input tokens. A human operator stayed in the loop and mostly stepped in at decision points; in one case the operator told the agents to look for machines that did not run the firm's own endpoint sensor.

Many entry routes were old weaknesses. CrowdStrike describes authentication keys from a 2022 directory export, probably never rotated, being used to forge an administrator token, and a VPN key that the operator said came from a 2024 email leak. Its recommendations include rotating signing keys and service credentials after any suspected exposure, enforcing SAML signature checks and patching internet-facing appliances.