CHINESE-SPEAKING attackers have leveraged Cairn, an AI-enabled pen-testing tool, to drive a large-scale automated attack spree that compromised hundreds of websites worldwide. NorForge, which coordinated disclosure with Reuters, identified three activity clusters on exposed dashboards, with two deemed malicious.
The operation appears to be financially motivated rather than state-sponsored, and it exposed hundreds of sensitive secrets detailing how automated agents were instructed to identify vulnerabilities and establish persistent access. At least one attack reportedly installed a Web shell on a commercial site, giving the attacker ongoing server access.
Victims publicly named by Reuters include John Deere, SharkNinja, and Cummins; Deere said remedial action was taken but there was no indication of sensitive data being accessed, SharkNinja reported the incident was contained and affected a standalone product-support site, and Cummins was examining a potential breach on a third-party vendor’s site.
NorForge found more than 1,100 attack projects spread across 14 Cairn dashboards, with AI agents coordinating multiple stages of attacks rather than a single operation. The campaigns targeted e-commerce and payment infrastructure, utilising vulnerabilities such as SQL injection, unsafe file uploads, insecure deserialization, authentication bypasses, and server-side template injection, while some instructions directed agents to deploy web shells.
Analysts warn that AI broadens the potential target base and accelerates attack tempo, enabling hundreds or thousands of attempts at low cost. Defence recommendations emphasise continuous discovery and testing of Internet-facing assets, tighter credential and key management, and security measures that automate remediation and patching, alongside a plan to take down hostile Cairn panels where possible.