Attackers used a customized version of the open-source Cairn AI penetration-testing framework in 25 operations against at least 19 organizations, according to Unit 42. The researchers identified unauthorized access or data retrieval at 10 organizations during August–September 2026.
The November 12 report covers semiconductor supply chains, defense and policy organizations, and civil society. Reported outcomes included persistent administrator accounts in a production system and access to more than one million personal records at one organization.
Human operators chose targets and goals and could supply hints during operations. Cairn handled planning and execution using Zhipu GLM and DeepSeek models.
Unit 42 tracks the activity as CL-STA-1238 and assesses a China-nexus intelligence-collection link with moderate confidence. That attribution remains the researchers’ assessment.
The exploited weaknesses included authentication bypasses, weak credentials and exposed cloud APIs. Unit 42 recommends identity hygiene, reducing exposed attack surfaces and patching software as relevant defenses against this activity.
