CL-STA-1238 Cairn campaign
China-nexus attackers used a customized Cairn AI pentest framework on Zhipu GLM and DeepSeek to run 25 autonomous operations against 19 organizations, reaching 10 (Unit 42).
What happened
Palo Alto Networks Unit 42 reported on Nov 12 2026 that Chinese-speaking attackers, tracked as CL-STA-1238 and assessed with moderate confidence as a China-nexus intelligence-collection effort, ran 25 operations against at least 19 organizations from August to September 2026 using Cairn, an open-source AI penetration-testing framework (the same tooling seen in the Cairn mass intrusion reported in October). Targets clustered in advanced semiconductors, the US defense industry and civil-society organizations. Unit 42 says Cairn confirmed unauthorized access or data retrieval at 10 of them; Cybersecurity Dive adds that the attackers obtained dozens of valid credentials, created administrator accounts in a production system and exposed more than a million personal records at one organization.
How it worked
The operators gave each task a one-line goal and a root domain, sometimes with intelligence gathered beforehand such as VPN gateway addresses and shared group passwords for a semiconductor equipment supplier. Cairn then planned and ran the operation itself: AI workers read and wrote a shared “blackboard” of confirmed facts and next steps, looping until the model judged the goal met. Operations ran from 20 minutes to over 50 hours. The attackers used a customized fork hosted in Hong Kong with a US staging node, and drove it with Zhipu GLM 5.2 and DeepSeek V4 Flash and Pro, escalating to V4 Pro for targets that resisted Flash. A skill pack absent from the upstream project added scanning and exploitation tools.
Cairn’s stock prompts already tell the model never to refuse. The attackers added a Chinese-language preamble asserting the engagement was authorized and an AUTHORIZATION.md file per operation; Unit 42 says neither model refused.
Why it matters
It is another case, alongside ARTEX campaigns, of open-source agent frameworks paired with low-cost open models letting a small team attack many organizations at once. Unit 42’s Andy Piazza told Cybersecurity Dive such tooling lets attackers hit more targets faster without learning the underlying technology.
Timeline
Every dated event in the wiki that involves CL-STA-1238 Cairn campaign, newest first.
- Google threat intelligence reports the first fully autonomous AI exploit pipeline in the wild (exploit_agent)
Google Threat Intelligence Group's quarterly AI threat tracker describes exploit_agent, found on an exposed staging server: wired to frontier and local LLMs, it diffs popular web apps before and after patches and runs the exploits it drafts automatically, apparently against French media, entertainment and education sites. The report also cites LLM-built exploits for the PaperCut zero-days, use of the CAIRN framework against US aerospace, and a North Korea-suspected group (UNC6836) that rigged Claude Code settings files to fetch a backdoor.
Palo Alto Networks Unit 42 says Chinese-speaking attackers it tracks as CL-STA-1238 ran 25 operations in Aug-Sep 2026 with a customized fork of the open-source Cairn pentest framework, driven by Zhipu GLM 5.2 and DeepSeek V4 Flash/Pro. A Chinese-language preamble told the models the work was authorized and forbade refusal; neither model refused. Unit 42 says Cairn confirmed access or data retrieval at 10 organizations.
Connections
CL-STA-1238 Cairn campaign and everything one link away. Drag to move, scroll to zoom, click a node to open it.Open in the full graph →
What links here
Incident Agent-driven mass intrusion via Cairn · exploit_agent framework · Microsoft agent-framework PyPI compromise
Product ARTEX
AI lab DeepSeek
Company Palo Alto Networks