A China-nexus hacking group that CrowdStrike tracks as Vapor Panda used the agentic AI framework ARTEX, run mainly on the Chinese DeepSeek model, against more than 200 organizations in at least 14 countries starting Oct. 17, 2026, CrowdStrike said in a report published Oct. 29. The activity came from a single ARTEX controller over an 11-day window, CrowdStrike said.
About 140 of those targets were compromised, CrowdStrike told CyberScoop. The company said roughly 70% of the targets were either confirmed compromised or showed evidence of a successful intrusion with moderate confidence. In its own report, CrowdStrike Intelligence said it confirmed post-exploitation activity at nine organizations, where the attackers deployed the BLOODALCHEMY backdoor and carried out credential spraying, network scans and reconnaissance.
Targets included foreign ministries in Latin America, Europe, Asia and the Middle East, and one European country’s foreign ministry, presidency and prime minister’s office. A separate engagement targeted 19 U.S.-based organizations in the pharmaceutical, semiconductor, defense, nuclear and financial sectors. The group also compromised telecom providers in the U.S., Belgium and the Philippines and a Middle East interior ministry, CrowdStrike said.
Agents running in parallel
CrowdStrike described the operation as an escalation from AI-assisted single intrusions to multiple simultaneous engagements steered by a human operator. Recovered transcripts showed an operator issuing instructions in Chinese, labeled “human instruction,” directing the agents to write web shells, extract domain administrator passwords and inventory endpoint sensor coverage. In one case the operator told the agents to find Windows machines without CrowdStrike’s Falcon sensor.
Requests came as often as every 10 seconds, CyberScoop reported. Adam Meyers, senior vice president of counter adversary operations at CrowdStrike, told the outlet that hands-on-keyboard hacking typically involved up to 40 commands per hour, but “with ARTEX, we could see it exceed 1,500 commands an hour.” In a single intrusion, the planning agents consumed on the order of 1.4 billion input tokens, CrowdStrike said.
“This is the first time that we’re seeing an agentic adversary nation-state threat actor at this scale,” Meyers told CyberScoop.

A second ARTEX campaign
CrowdStrike said this campaign is separate from the ARTEX activity Microsoft attributed earlier this month to Storm-0940, which The Vanguard USA covered on Oct. 22 in connection with the ConnectWise Cloud Backup attack. The two campaigns likely indicate that the framework is quickly gaining popularity among multiple China-nexus groups, CrowdStrike said.
Vapor Panda has targeted government, think tank, energy and extractive organizations in Europe, North America, Oceania and the UK since 2021, according to CrowdStrike. This campaign is the group’s first observed use of AI-driven intrusion tooling and broadens its targeting into the pharmaceutical, semiconductor, nuclear and defense sectors. The campaign remains active, CyberScoop reported.
CrowdStrike observed roughly 45 agentic adversary attacks in September and 126 so far in October, according to CyberScoop. Meyers said organizations need visibility into their endpoints, identities and cross-domain activity. “It’s not hopeless, but you’ve got to take action,” he said.


