Trusence Every claim has a source
Last updated 17 November 2026 Search Türkçe
Security

AI agents were used to automate breaches at John Deere, SharkNinja and others

The case shows how a small amount of human direction can scale attacks across many websites and countries.

Security researchers at Norforge say a hacker used exposed Cairn AI-agent instances to automate intrusions against more than 100 websites, showing how attacker-controlled agent workflows can scale breaches with little direct supervision. Reuters said the victims included companies such as John Deere, SharkNinja, and Cummins, with targets spread across the US, Canada, Australia, and Japan. Norforge described the most active operator as using fewer than 100 Chinese characters to direct the agents, and Reuters said successful breaches took about 23 hours while the agents were also seen attempting more than 1,200 websites. John Deere said there was no indication sensitive information was accessed, SharkNinja said the attack was contained to a product-support site, and Cummins said it was investigating a possible cyberattack on a vendor-run website.

Why it matters

This matters because it shows attacker-controlled AI workflows can be used to reach many sites with little direct supervision, including large companies in several countries. The reported pace and scale make it harder for target organisations to rely on manual response alone, and they may need to treat agent-driven abuse as a broader class of attack rather than a one-off intrusion.

Sources · 2