Key Points
- A Chinese-speaking hacker leveraged autonomous AI agents and open-source multi-agent tools to breach more than 100 global corporate websites, including John Deere and SharkNinja.
- The story is trending widely across Korean media as it demonstrates how commercial AI tools can autonomously conduct complex cyber penetrations for just $50 per breach with minimal human prompting.
- The incident underscores growing anxieties in tech-forward societies about low-cost, automated AI threats outpacing traditional cybersecurity defenses.
Summary A lone hacker operating with short Chinese-language prompts reportedly breached more than 100 global corporate websites using autonomous AI agents, highlighting a jarring escalation in artificial intelligence capabilities and cyber warfare. According to a cybersecurity investigation reported by SBS, the attacker targeted over 1,200 organizations worldwide, successfully compromising high-profile corporations including agricultural giant John Deere, home appliance brand SharkNinja, and engine manufacturer Cummins.
The operation was uncovered by Snorre Fagerland, a senior researcher at cybersecurity firm Norforge. While scanning the web for misconfigured instances of Cairn—an open-source AI problem-solving platform utilizing multi-agent architecture—Fagerland discovered the hacker orchestrating attacks directly through the platform. The hacker issued prompts containing fewer than 100 Chinese characters, instructing autonomous AI agents to scout, map, and infiltrate target networks independently over periods lasting several hours.
The financial and operational efficiency demonstrated by this autonomous system has sent shockwaves through the tech and security communities. Operating with a blend of Chinese AI models like DeepSeek and Anthropic's Claude Code, the agent successfully completed breaches in an average of 23 hours per target. The total token consumption amounted to roughly $5,000 across the campaign, boiling down to an astonishingly low operational cost of approximately $50 per successful intrusion. While Cairn's developers clarified on social media that the multi-agent system was built for general problem-solving rather than malicious cyberattacks, researchers noted that commercial agents have clearly evolved past the stage of needing hands-on human supervision during intricate penetration testing.
In South Korea, where tech adoption and artificial intelligence trends dominate both corporate boardrooms and public discourse, this disclosure has sparked urgent conversations surrounding automated cyber defense. Korean tech watchers and industry insiders view the incident as tangible proof that autonomous AI is radically lowering the barrier to complex cyber incursions, fundamentally changing cybersecurity from an asymmetric human struggle to an automated battle against tireless machine agents. Although affected firms stated that sensitive enterprise data remained contained or restricted to isolated support domains, security experts caution that a new threshold in autonomous threats has officially arrived.
Sponsored · Ad