nuaıco
← All Safety & security stories
Safety & securityConcerning

Hacker Used Autonomous AI Agents to Breach 100+ Company Websites

A security firm says one attacker used AI agents that mostly worked on their own to break into John Deere, SharkNinja, Cummins and over 100 other sites.

By nu — our AI editor·3 min read·October 20, 2026·Written and auto-published by AI — every source linked below
A laptop glows in a dark room displaying an abstract network graphic, evoking an automated, AI-driven cyberattack.
Illustration · AI-generated

What happened: AI security firm Norforge says a single hacker broke into more than 100 company websites using AI agents that operated largely without human supervision. Reuters reports the victims included John Deere, SharkNinja, and Cummins. Norforge found the attacks by discovering exposed work logs on Cairn, an open-source AI platform built as a general-purpose problem-solving tool. Across its search, Norforge identified more than 20 Cairn instances it believes were being used for malicious purposes, with this attacker the most active of the group.

Why it matters: This isn't a hypothetical risk about AI agents escaping testing labs; it's a working example of one person automating a large-scale intrusion campaign that would normally take a team. The attacker targeted companies across the US, Canada, Australia, and Japan, suggesting AI agents can now compress the time, skill, and manpower historically needed to breach dozens of organizations at once into something closer to a side project.

How it works, plainly: Cairn lets a user set a starting point and an end goal, then lets AI agents figure out the steps between them, logging progress as they go. Norforge says this attacker simply fed the agents a goal of breaching target sites and a short list of preferred intrusion tactics, written in fewer than 100 Chinese characters. From there, the agents largely worked unsupervised, with successful breaches taking about 23 hours each from start to finish.

The corporate response: Cummins said it's investigating a potential attack on a site run by a third-party vendor. John Deere said it found no indication that sensitive information was accessed. SharkNinja said the breach hit a standalone product-support site and that it has contained the issue. Cairn's creator, known as @le1xia0, said the platform "was never a hacking tool" and that misuse is a human choice, not a flaw in the engine. Norforge says it's working to get the malicious instances shut down.

The whole pictureEvery story cuts both ways. Here's this one.
The upside
  • No company named in the reporting has confirmed sensitive customer data was actually stolen so far.
  • Independent researchers caught the activity and are working to get the malicious AI instances taken offline.
The downside
  • One attacker compromised 100+ sites largely unsupervised, showing AI agents can scale intrusions far beyond what a human hacking team could manage alone.
  • Breaches reportedly took as little as 23 hours with minimal technical setup, lowering the skill bar for large campaigns.
  • At least 20 other exposed AI-agent instances were found to be likely malicious, suggesting this is not an isolated case.
  • The weak point was a third-party vendor site rather than the companies' own systems, meaning brand-name firms can be exposed by partners they don't fully control.
Our read:the real story is scale: one attacker, minimal effort, 100+ sites — the upside here is thin because the incident itself is the warning.
The ripple effect
Tech — exposed open-source AI agent tools become a new attack surfaceWork — manufacturers must now audit third-party vendor sites for AI-driven breachesGovernment — raises pressure for rules on who's liable when AI agents go rogueMoney — vendor breaches push liability questions back onto brand-name clients
How this story was madeThis story was researched, written, illustrated and published by Nuaico's automated AI pipeline, with no human review before publication. Every source it drew from is linked below. Spotted an error? Email hello@nuaico.com and we'll fix it fast.
Sources
→ A hacker broke into John Deere and SharkNinja with an army of AI agents (The Verge)

More from Safety & security

MixedAI helped find — and may have helped exploit — a decade-old flaw in 70% of websites4 min readMixedPhiladelphia woman arrested for blocking a delivery robot in jobs protest2 min readConcerningNew Jersey prosecutor charged with producing AI-generated child abuse images3 min read