AI Agents Compromise 100 Company Websites in Global Campaign

A Chinese-speaking hacker utilized autonomous AI agents to compromise more than 100 company websites, including those of Deere, SharkNinja, and Cummins. The campaign, which began September 25, targeted e-commerce sites in the United States, Canada, Australia, and Japan, with the attacker attempting over 1,200 break-ins.

Snorre Fagerland, a researcher at NorForge, discovered the activity by monitoring exposed instances of Cairn, an open-source AI problem-solving platform. The hacker combined DeepSeek and Anthropic's Claude Code to execute the attacks, spending approximately $5,000 in tokens. Fagerland observed that the AI agents operated without human intervention to identify infrastructure weaknesses and execute malicious commands.

Deere reported taking remediation actions and found no evidence that sensitive data was accessed. SharkNinja stated the breach was limited to a standalone product support site, while Cummins is investigating a potential attack on a third-party vendor's website. The author of the Cairn platform, le1xia0, stated the tool was not designed for hacking attacks.