Chinese-speaking actor uses Cairn AI agents with DeepSeek and Claude Code to breach 100+ company websites for about $50 each
A Chinese-speaking threat actor used AI agents coordinated through the open-source Cairn platform to attempt intrusions against more than 1,200 websites, reportedly compromising sites of over 100 companies, including Deere, SharkNinja and Cummins. Researcher Snorre Fagerland of NorForge found the activity while monitoring internet-exposed Cairn instances used for malicious purposes, and followed part of it from 2 October. Per Reuters, the campaign began around 25 September and targeted sites in the United States, Canada, Australia and Japan, using short Chinese-language prompts (under 100 characters). Fagerland estimated the cost at about US$5,000 in AI credits, with each successful intrusion taking roughly 23 hours and costing about US$50. The operation apparently combined DeepSeek with Anthropic's Claude Code.
Technical Details
- The agents autonomously searched for vulnerabilities and attempted intrusions with little human intervention.
- In one case, after finding no flaws in a large company's public application, the agents pivoted to analyzing digital (DNS-type) records to map infrastructure tied to the target. This led them to an internet-exposed control panel with a data-processing flaw, which they reported exploiting to make the server execute malicious commands.
- The article does not name the vulnerability or give a CVE.
- Fagerland also saw other criminal uses of Cairn instances: one actor stole a large volume of credit card data, and another tried to use agents to rob a cryptocurrency exchange.
- The article publishes no IOCs.
Remediation
The article gives no specific remediation guidance. Reasonable priorities are to reduce exposure of administrative panels on the internet and to patch data-processing flaws in exposed services, since AI agents can pivot to secondary infrastructure when the main application is hardened.
Linha do tempo
-
Chinese-speaking actor uses Cairn AI agents with DeepSeek and Claude Code to breach 100+ company websites for about $50 each
ler a análise desta fonte ocultar
A Chinese-speaking threat actor used AI agents coordinated through the open-source Cairn platform to attempt intrusions against more than 1,200 websites, reportedly compromising sites of over 100 companies, including Deere, SharkNinja and Cummins. Researcher Snorre Fagerland of NorForge found the activity while monitoring internet-exposed Cairn instances used for malicious purposes, and followed part of it from 2 October. Per Reuters, the campaign began around 25 September and targeted sites in the United States, Canada, Australia and Japan, using short Chinese-language prompts (under 100 characters). Fagerland estimated the cost at about US$5,000 in AI credits, with each successful intrusion taking roughly 23 hours and costing about US$50. The operation apparently combined DeepSeek with Anthropic's Claude Code.
Technical Details
- The agents autonomously searched for vulnerabilities and attempted intrusions with little human intervention.
- In one case, after finding no flaws in a large company's public application, the agents pivoted to analyzing digital (DNS-type) records to map infrastructure tied to the target. This led them to an internet-exposed control panel with a data-processing flaw, which they reported exploiting to make the server execute malicious commands.
- The article does not name the vulnerability or give a CVE.
- Fagerland also saw other criminal uses of Cairn instances: one actor stole a large volume of credit card data, and another tried to use agents to rob a cryptocurrency exchange.
- The article publishes no IOCs.
Remediation
The article gives no specific remediation guidance. Reasonable priorities are to reduce exposure of administrative panels on the internet and to patch data-processing flaws in exposed services, since AI agents can pivot to secondary infrastructure when the main application is hardened.