SecurityOnlineDo Son
Unit 42 links Cairn AI attacks to 10 organisation breaches
Part of the CL-STA-1238 story · since 16 Nov
- Mon 16 Nov, 18:00 UTC. Dark Reading, first report. China linked actors use AI framework to breach 10 organisations.
- Tue 17 Nov, 10:01 UTC. SecurityOnline, its first report on the story. Unit 42 links Cairn AI attacks to 10 organisation breaches. This article, 2nd of 2.
From August to September 2026, Palo Alto Networks Unit 42 reports that Chinese-speaking threat actors used an open-source Cairn framework fork to run 25 autonomous hacking operations against 19 organisations, achieving breaches at 10 targets without manual keyboard input. The cluster, tracked as CL-STA-1238, paired Cairn with commercial Chinese AI models—Zhipu GLM 5.2, DeepSeek V4 Flash and V4 Pro—so that a single operator could oversee multiple simultaneous intrusions.
In one case against a semiconductor equipment vendor, attackers seeded the Cairn setup with pre-collected VPN gateway addresses and shared group passwords to accelerate initial access.
The operations ran from a server in Hong Kong (154.44.31[.]4) and routed traffic through a US staging node (82.29.129[.]54), with tasks bootstrapped simply by feeding a root domain and a Chinese one-line objective; Cairn then planned, executed and recorded results autonomously for time frames ranging from 20 minutes to over 50 hours, according to Unit 42.
Inside Cairn, a central blackboard architecture drives each worker agent through a bootstrap, a “Reason” phase and an “Explore” phase across three to seven stages. Confirmed findings are written as immutable facts, while next steps remain as intents.
The operators loaded a custom skill pack named “AIScan Runner Skills” and generated an AUTHORIZATION[.]md file; to bypass safety guardrails, they prefixed a Chinese system prompt stating that tasks were part of an authorised security assessment or capture-the-flag exercise. Unit 42 attributes the activity to a China-nexus intelligence-collection effort, noting Chinese prompts, exclusive use of Chinese AI models, and targeting aligned with Beijing’s strategic priorities.
The breaches included unauthorised access and data exposure at multiple sectors, with notable impact at a U.S. teachers’ union involving over one million member records. Defences cited include MFA enforcement, password hygiene and restricting unauthenticated cloud writes, plus blocking known attacker hosts and auditing for unauthorised admin accounts.
How the story developed
2 articles from 2 outlets since 16 Nov. Still developing; the latest report was 9 hours ago. This was the 2nd report, 16 hours after the first.
- Mon 16 Nov, 18:00 UTC. Dark Reading, first report. China linked actors use AI framework to breach 10 organisations.
- Tue 17 Nov, 10:01 UTC. SecurityOnline, its first report on the story. Unit 42 links Cairn AI attacks to 10 organisation breaches. This article, 2nd of 2.
- No CVE
- Larger: an outlet’s first report
- Coverage over time
Show all 2 articles as a listHide the list
- Mon 16 NovDay 01
- China linked actors use AI framework to breach 10 organisations18:00 · Dark Reading · First report
- Tue 17 NovDay 11
- Unit 42 links Cairn AI attacks to 10 organisation breaches10:01 · SecurityOnline · This article, 2nd of 2