100 Global Giants Hacked in Unprecedented 50-Per-Breach Wave
A dangerous paradigm shift has shaken the cybersecurity landscape after a Chinese-speaking cybercriminal orchestrated automated cyberattacks against more than 100 major corporations worldwide—including agricultural machinery giant Deere, home appliance maker SharkNinja, and industrial engine manufacturer Cummins—by deploying fully autonomous Artificial Intelligence agents without writing complex exploit code. Monitored in real time by threat researchers, the massive campaign demonstrated how generative and agentic AI technologies can be weaponized with minimal human direction, delegating network scanning, vulnerability discovery, and server infiltration entirely to algorithmic agents. The breach underscores the emergence of an asymmetric cyber threat vector where individual actors can inflict widespread commercial disruption on global enterprises with virtually zero manual technical effort.
Inside the Autonomous Assault: 100-Word Prompts and $50 Exploitation Runs
The live intrusion was uncovered by Snorre Fagerland, principal threat researcher at cybersecurity firm Norforge, who detected nearly two dozen instances of an open-source multi-agent framework named Cairn being mobilized for malicious operations around October 2, 2026. The technical trail revealed that the threat actor initiated automated reconnaissance on September 25 across roughly 1,200 commercial and e-commerce targets spanning the United States, Canada, Australia, and Japan. Rather than authoring bespoke exploit payloads, the perpetrator provided the autonomous agents with rudimentary guidance consisting of fewer than 100 Chinese words, leveraging underlying AI models from China’s DeepSeek alongside developer toolkits like Anthropic's code utilities. The autonomous software conducted extensive reconnaissance for up to 23 hours per target, independently scouring secondary digital records when main portals proved secure, locating exposed control panels, and exploiting hidden coding flaws to seize server-level access. In total, the cybercriminal expended approximately $5,000 worth of computational API tokens across the entire operation, translating into an astonishingly low operational cost of roughly $50 per breached company.
Corporate Damage Control and the Escalating Danger of Unsupervised AI Weaponization
Following the exposure of the breaches, the affected multinational corporations initiated immediate incident response protocols to mitigate operational exposure. Deere confirmed that prompt remediation measures were deployed across its digital perimeter, emphasizing that internal forensics detected no evidence of sensitive customer or proprietary data theft. SharkNinja reported that the intrusion was isolated to a peripheral domain hosting legacy product support manuals and that the underlying vulnerability has been contained, while Cummins launched an in-depth investigation into an external corporate site operated by a third-party vendor. While Cairn framework creator le1xia0 clarified on social media that the multi-agent software was designed exclusively for administrative automation rather than malicious hacking, security analysts caution that unsupervised agentic architectures are evolving beyond human oversight. With comparable automated AI-driven campaigns recently identified accessing financial repositories and credit card databases across South Korea, Japan, and Western economies, the episode confirms that modern cyber warfare has entered an era where lone operators can utilize commercial AI tools to launch enterprise-grade cyber offensives at global scale.
