A single hacker used autonomous AI agents to target more than 1,200 websites and break into over 100 of them, according to a Reuters investigation. The campaign shows how one person can automate a large hacking operation with limited hands-on work.
1,200 websites targeted
NorForge cybersecurity researcher Snorre Fagerland observed the campaign through exposed control panels for Cairn, an open-source AI problem-solving platform. Reuters identified Deere, SharkNinja and Cummins among the companies linked to the activity. The full list of affected sites and the amount of data taken remain unclear.
According to Fagerland, the agents typically spent about 23 hours probing each successful target. He estimated that the campaign used roughly $5,000 in AI tokens, or about $50 for each successful intrusion. Reuters reported that the activity appeared to use a mixture of DeepSeek and Anthropic’s Claude Code, although the exact model setup was not independently confirmed.
Cairn automated the campaign
Cairn lets an operator set a target and goal while multiple agents share discoveries and continue working through a problem. In this campaign, the hacker allowed the agents to search for weaknesses and attempt intrusions with little further direction. NorForge documented more than 20 exposed Cairn panels being used in suspected criminal activity in its technical report.
The companies linked to the activity reported different levels of impact. Deere said it found no indication that sensitive information was accessed. SharkNinja said a separate product-support website was affected and the incident was contained. Cummins said it was investigating a possible attack involving a website operated by a third-party vendor.
Microsoft details agent safeguards
The investigation comes as companies give AI agents broader access to business systems. Microsoft’s official security post describes assigning agents identities, containers and policy-based permissions, along with continuous monitoring. Microsoft also describes “Guardian Agents” that can supervise other agents as risk changes.
Microsoft says its Project Perception system is designed to monitor agent activity, assess changing risks and adjust protections in real time. Those controls address the same type of autonomous activity highlighted by the Reuters investigation.
References
Fagerland, S. (2026, October 19). Inside Cairn: Open windows into massive AI-driven cyber operations. NorForge. https://norforge.com/blog/inside_cairn
Gallot, H. (2026, October 19). Securing the agentic insider. Microsoft Official Blog. https://blogs.microsoft.com/blog/2026/10/19/securing-the-agentic-insider/
Satter, R. (2026, October 19). How one hacker used AI agents and a few thousand dollars to break into scores of sites. Reuters. https://www.reuters.com/legal/government/how-one-hacker-used-ai-agents-few-thousand-dollars-break-into-scores-sites-2026-10-19/