---
title: Unit 42: a China-nexus actor pointed the open-source Cairn AI pentest framework, running on GLM 5.2 and DeepSeek V4, at 19 organizations and confirmed access at 10 | ctxsec
url: https://ctxsec.com/items/unit42-cl-sta-1238-cairn-autonomous-ai-pentest-framework-19-orgs-semiconductor-defense-civil-society
description: "Unit 42 says Chinese-speaking attackers (tracked as CL-STA-1238) ran a customized fork of Cairn, an open-source framework where AI 'agent workers' share a blackboard of facts and intents, in 25 operations against at least 19 organizations in August and September 2026: seven in the semiconductor and advanced-manufacturing supply chain, three in defense and defense policy, and four human-rights and civil-society groups. Each operation got a one-line goal such as obtaining webshells, SSH, source code, keys or credentials, ran 20 minutes to over 50 hours, and used Zhipu GLM 5.2 (13 operations) or DeepSeek V4 Flash/Pro (12), plus a custom 'AIScan Runner Skills' pack. A Chinese-language preamble told the model every task was authorized and never to refuse; neither model refused. Cairn confirmed access or data retrieval at 10 organizations, including over a million member records at a US teachers' union, an overwritten standards-ballot document, and rogue admin accounts at a semiconductor firm. Unit 42 stresses the flaws were known ones: auth bypasses, IDOR, weak credentials, exposed cloud APIs and old CVEs."
status: "vendor research (Unit 42, Nov 12, 2026); China-nexus attribution at moderate confidence"
published: 2026-11-14
---

> Unit 42 says Chinese-speaking attackers (tracked as CL-STA-1238) ran a customized fork of Cairn, an open-source framework where AI 'agent workers' share a blackboard of facts and intents, in 25 operations against at least 19 organizations in August and September 2026: seven in the semiconductor and advanced-manufacturing supply chain, three in defense and defense policy, and four human-rights and civil-society groups. Each operation got a one-line goal such as obtaining webshells, SSH, source code, keys or credentials, ran 20 minutes to over 50 hours, and used Zhipu GLM 5.2 (13 operations) or DeepSeek V4 Flash/Pro (12), plus a custom 'AIScan Runner Skills' pack. A Chinese-language preamble told the model every task was authorized and never to refuse; neither model refused. Cairn confirmed access or data retrieval at 10 organizations, including over a million member records at a US teachers' union, an overwritten standards-ballot document, and rogue admin accounts at a semiconductor firm. Unit 42 stresses the flaws were known ones: auth bypasses, IDOR, weak credentials, exposed cloud APIs and old CVEs.

[← AI](/ai) · [markdown](/items/unit42-cl-sta-1238-cairn-autonomous-ai-pentest-framework-19-orgs-semiconductor-defense-civil-society.md)

vendor research (Unit 42, Nov 12, 2026); China-nexus attribution at moderate confidence

2026-11-14 7:40 AM ET

### Unit 42: a China-nexus actor pointed the open-source Cairn AI pentest framework, running on GLM 5.2 and DeepSeek V4, at 19 organizations and confirmed access at 10

- **IOCs:** 154.44.31[.]4 (Cairn controller, Hong Kong, AS979) and 82.29.129[.]54 (staging and attack egress, US, AS142036).
- **Why it matters:** one operator or small team ran simultaneous intrusions against 19 organizations in two weeks; volume rises, complexity doesn't.
- **Defend:** identity hygiene (no shared or default passwords, MFA on VPN and partner portals), close open registration and account-creation endpoints, check authorization on every API object, and patch old public CVEs.

Sources: [Unit 42: Attackers Use Autonomous AI to Target Semiconductor, Defense and Civil Society Sectors (Nov 12, 2026)](https://unit42.paloaltonetworks.com/cairn-framework-ai-autonomous-attacks/)
