N環球 
Hacker used autonomous AI agents to breach websites of 100+ companies, Reuters reveals
Cybersecurity firm NorForge reports that a Chinese-speaking hacker used fewer than 100 Chinese characters of instructions to have autonomous AI agents scan and breach more than 1,200 websites across the US, Canada, Australia and Japan — over 100 companies were compromised.
Why it matters
This is the first well-documented case of large-scale "AI-automated hacking": the attacker barely lifted a finger while commercially available AI agents independently carried out full attack chains, from reconnaissance to intrusion. The researchers say it shows how AI massively amplifies a lone attacker's capabilities — "a whole new world."
Full report
Reuters, citing a report by Snorre Fagerland, principal threat researcher at cybersecurity firm NorForge, says a Chinese-speaking hacker used the open-source AI agent platform Cairn to direct agents to autonomously scan and break into target websites. Fagerland found about 20 maliciously used Cairn instances in early October while scanning the internet; their accidentally exposed control panels let researchers watch the attacks in real time.
The campaign began on September 25, targeting e-commerce websites in the United States, Canada, Australia and Japan. By last week, the agents had attempted intrusions against more than 1,200 websites and successfully breached websites belonging to over 100 companies, including agricultural machinery giant John Deere, appliance maker SharkNinja and engine manufacturer Cummins.
The attacker's instructions ran to fewer than 100 Chinese characters, directing the agents to find ways to run code on each site and then plant an uploader web shell. A successful breach took about 23 hours on average; the whole campaign is estimated to have cost only about $5,000 in AI tokens — roughly $50 per successful intrusion. The models reportedly used were a mix of China's DeepSeek and Anthropic's Claude Code.
John Deere told Reuters it had taken necessary remediation actions and had no indication that sensitive information was accessed. SharkNinja said the incident affected a standalone website carrying only product-support information and that the threat had been contained. The author of Cairn (posting as "le1xia0") responded that the tool was not designed for hacking attacks and that harmful use was not their intention.
💬 Comments
No comments yet. Be the first!