Single hacker used AI agents to breach John Deere and SharkNinja

By Splitfeed EditorialLast updated Oct 19, 2026, 15:01 UTC

On October 19, 2026, Reuters published an exclusive investigation, based on findings from AI security firm Norforge, detailing how a single unidentified hacker breached more than 100 websites using AI agents that operated with little human direction. Named victims include John Deere, SharkNinja, and Cummins. Norforge documented the campaign without publicly identifying all targets. The cost per intrusion ran to approximately $50, with the hacker's total spending reaching only a few thousand dollars across the entire operation.

The figures are what make this incident significant beyond a standard breach disclosure. Large-scale intrusion campaigns have historically required either substantial resources, technical expertise, or both. This case challenges that assumption. According to Raphael Satter, who reported on the findings, the hacker provided short, generic instructions running to fewer than 100 Chinese characters, and the AI agents then conducted complex intrusions with no further human input in some instances. What is genuinely in dispute is how representative this case is, whether existing defenses can adapt quickly enough, and whether AI tool providers bear any responsibility for closing off these use cases.

The Anti-AI camp has responded with notable alarm. Raphael Satter described the operation as "almost entirely autonomous hacking," and Bruno J. Navarro noted that "the hacker did almost none of the work," framing the AI agents as the primary actors rather than mere tools. Both reactions center on the delegation of attack complexity to automated systems as a threshold that has now been crossed at negligible cost. The Pro-AI camp has not yet published reactions to this story; typically that camp would argue that the same autonomous capabilities accelerate defensive security work and that restricting AI tools would disadvantage defenders more than attackers. The Middle Ground camp has also not weighed in yet; that camp would typically call for targeted safeguards on agentic systems without broader restrictions, emphasizing that the problem lies in deployment guardrails rather than the technology itself.

The details most worth watching are whether any of the named companies disclose the scope of data accessed, whether Norforge or law enforcement identifies the hacker, and whether AI tool providers announce changes to usage policies or technical controls in response to the documented attack pattern.