CrowdStrike: China-linked hackers used AI agents running mainly on DeepSeek against 200+ organisations in 11 days

The cybersecurity company CrowdStrike reported on 29 October 2026 that a China-linked espionage group it tracks as VAPOR PANDA used ARTEX, an open-source framework of AI agents for penetration testing, to attack many targets at the same time.

What CrowdStrike found

• From 17 October, a single ARTEX controller was aimed at more than 200 organisations in at least 14 countries over 11 days: foreign ministries in Latin America, Europe, Asia and the Middle East, US semiconductor, defence and think-tank organisations, and telecom operators in the US, Belgium and the Philippines.
• The agents ran mainly on the Chinese DeepSeek model, plus a second subscription-based commercial model that CrowdStrike could not identify. In one intrusion the planning agents used about 1.4 billion input tokens.
• CrowdStrike confirmed follow-on activity inside nine organisations, where the group installed the BLOODALCHEMY backdoor, and saw password spraying against at least 12 more.
• A human stayed in the loop: recovered transcripts show an operator giving instructions in Chinese, for example to look for Windows machines without CrowdStrike’s own sensor.

A CrowdStrike executive told CyberScoop that human hackers typically type up to about 40 commands an hour, while with ARTEX activity exceeded 1,500 commands an hour.

Caveats

The figures come from CrowdStrike’s own investigation and have not been independently verified. “200 targeted” does not mean 200 breached: the company confirmed post-exploitation at nine. China usually denies such accusations; NBC News and CyberScoop did not report a Chinese government response in their coverage.