CrowdStrike says China-backed group used DeepSeek-based AI against 200 organizations

CrowdStrike says VAPOR PANDA targeted more than 200 organizations in at least 14 countries and compromised about 140 sensitive targets during an 11-day period.[1] The company told CyberScoop that about 70% of the group’s targets were confirmed compromised or had evidence suggesting an intrusion with moderate confidence.[1] Those figures describe different measures: the reported target count, the number of sensitive targets compromised, and a broader estimate that includes moderate-confidence evidence.[1] The evidence does not establish that every targeted organization was breached.

CrowdStrike’s report describes ARTEX as an automated agentic AI penetration-testing framework, run primarily on DeepSeek and a second commercial model.[1][2] The group pursued multiple attack paths concurrently, and requests occurred as often as every 10 seconds.[1] CrowdStrike’s Meyers said ARTEX activity could exceed 1,500 commands an hour, compared with up to 40 commands per hour for the hands-on-keyboard activity the company had tracked over 15 years.[1] NBC News reported that AI enabled a scale that previously required substantially more human effort.[3]

CrowdStrike identified foreign ministries, a European country’s presidency and prime minister’s office, and defense and intelligence agencies among the targets and confirmed victims.[1] Nineteen U.S.-based organizations in pharmaceuticals, semiconductors, defense, nuclear and financial services were also among the targets or confirmed victims.[1] Telecom providers in the United States, Belgium and the Philippines were identified as targets or confirmed victims.[1] CrowdStrike said the active campaign targeted think tanks and government agencies in defense, justice, finance and diplomacy.[1]

CrowdStrike said the VAPOR PANDA campaign remained active and described its target sectors and geographies as ones Beijing considers economically or politically important.[1] Meyers called it the first agentic, nation-state adversary campaign of this scale that CrowdStrike had seen, and said malicious use of agentic AI was growing month over month at almost threefold.[1] CrowdStrike observed roughly 45 agentic-adversary attacks in September and 126 in October, with the October count still rising.[1] CyberScoop noted this campaign was separate from ARTEX activity Microsoft Threat Intelligence attributed to Storm-0940.[1]