An investigation by cybersecurity firm NorForge has uncovered more than 20 publicly exposed artificial intelligence control panels linked to criminal hacking campaigns, revealing how inexpensive AI tools are being used to target e-commerce platforms, payment systems, and cryptocurrency exchanges.
The report, published Monday, follows a warning from Anthropic in November 2025 that a state-backed group had used its Claude Code tool to attack roughly 30 targets, with AI performing 80% to 90% of the work and humans intervening at only four to six key points. By 2026, NorForge says similar capabilities are available to ordinary criminals at far lower cost.
Researcher Snorre Fagerland began his investigation after Gambit Security identified open-source AI tools Hermes, Strix, and Cairn. Cairn is a free platform that lets users assign a target and a goal, after which AI agents search for weaknesses and attempt intrusions, sometimes for hours. Using Shodan and Censys, Fagerland found more than 20 exposed Cairn panels believed to have been used for crime. The panels show attack targets, discovered vulnerabilities, stolen passwords, and access keys.
One campaign cost an estimated $5,000 in AI fees, roughly $50 per successful intrusion, using a combination of DeepSeek technology and Anthropic Claude Code. The busiest operator gave AI agents instructions in fewer than 100 characters and targeted more than 100 companies, including Deere, SharkNinja, and Cummins. From September 25, agents attempted to break into more than 1,200 retail websites in the United States, Canada, Australia, and Japan, with each successful intrusion taking about 23 hours of automated probing.
Other operators pursued different goals. One stole a large volume of credit card data, while another attempted to rob a cryptocurrency exchange using AI agents. NorForge found that the largest cluster, designated AIA0001, included 14 Chinese-language panels running more than 1,100 attack projects, most aimed at online stores and payment systems. The security firm is withholding panel addresses and victim data while working with affected parties to remove the exposed infrastructure.
Cairn's creator, posting as le1xia0 on X, said the tool was an innovative multi-agent architecture and was not designed for hacking attacks. The findings also come as regulators examine AI safety risks. The Federal Trade Commission sent legal demands to OpenAI and Anthropic on October 14, and Anthropic has floated draft legislation that would allow the government to block AI models posing at least $1 billion in damage risk.
Deere said it has taken remediation steps and has no indication that sensitive information was accessed. SharkNinja said the incident affected a separate product-support site and had been contained, while Cummins said it was investigating a possible attack involving a third-party vendor site. No confirmed financial losses have been disclosed, and NorForge has not established the full extent of data theft or stolen funds.
