CrowdStrike reported that a Chinese espionage group it tracks as Vapor Panda used ARTEX, a multi-agent AI penetration-testing framework developed in China and first released as open source, in attacks on more than 200 organizations in 14 countries. The operation began on October 17 and targeted government ministries, telecommunications providers, and organizations in the pharmaceutical, semiconductor, defense, nuclear and financial sectors. According to BleepingComputer, at least nine organizations were compromised, and password spraying and credential scanning were carried out against 12 other targets.
Confirmed victims include telecom providers in the US and Belgium, government bodies in Greece, the Middle East and Latin America, and US pharmaceutical firms. According to the researchers, the attackers used DeepSeek and a second, unidentified commercial language model to coordinate planning and worker agents across several intrusions at once, while the operators gave instructions in Chinese and stepped in at decision points. CrowdStrike estimates that in one intrusion the agents consumed more than 1.4 billion input tokens, and calls the campaign an escalation: from single intrusions to multiple simultaneous attacks directed by a human operator.
Separately, the Dutch Institute for Vulnerability Disclosure (DIVD) said the tool that attacked its network in September was ARTEX and that it acted autonomously, going from initial access to root privileges in 12 minutes. According to the report, the agents were told to skip machines protected by CrowdStrike's endpoint product and focused on unprotected hosts, and in the attack on a Greek government body they used authentication keys from an old mail-server export. What to do: make sure every server and workstation is covered by endpoint protection and patched, and rotate old keys and credentials.