Single Hacker Breaches 100+ Companies Using AI Agents

A Chinese-speaking hacker used autonomous AI agents to breach websites belonging to more than 100 companies, including Deere, SharkNinja, and Cummins, according to a Reuters investigation published October 19, 2026. The attacker spent only $5,000 on AI tokens to compromise over 1,200 sites in a campaign that began September 25, with individual successful intrusions taking approximately 23 hours each.

Security researchers uncovered a large-scale cyberattack campaign demonstrating how artificial intelligence agents have dramatically lowered the barriers to conducting sophisticated cybercrimes. A single hacker, identified as Chinese-speaking, delegated nearly all aspects of website intrusions to commercially available AI agents, requiring minimal human involvement beyond providing high-level objectives in fewer than 100 characters of instruction.

The campaign targeted over 1,200 websites across four countries, successfully breaching more than 100 company sites. Among the victims were major industrial manufacturers including John Deere, SharkNinja, and Cummins, though the extent of data exposure varied across organizations. According to NorForge researcher Snorre Fagerland, the operation achieved this scale and scope for an estimated $5,000 in AI token costs, averaging approximately $50 per successful intrusion. The targeted companies' responses indicated varying levels of impact, with Deere reporting remedial action and no sensitive data access, while SharkNinja confirmed the intrusion affected only a standalone product-support website.

What distinguishes this attack from traditional cybercriminal operations is the relationship between cost, effort, and output. Successful intrusions typically required only 23 hours of automated agent work, with the attacker providing minimal direct oversight. Security researchers note that the incident represents a fundamental shift in the economics of cybercrime, as agentic AI has collapsed the skill and financial barriers that previously separated casual attackers from capable ones. The hacker's approach, weaponizing rather than exploiting traditional software vulnerabilities, highlights how autonomous AI systems can be redirected toward malicious purposes at scale.