How one hacker used AI agents and a few thousand dollars to break into scores of sites

Reuters | Raphael Satter

Chinese-speaking hackers deployed autonomous artificial intelligence agents to breach more than 100 corporate websites in a low-cost campaign tracked by cybersecurity researcher Snorre Fagerland. The intruder used the open-source platform Cairn, directing AI tools with fewer than 100 characters of instructions to execute sprawling intrusions against e-commerce networks in the US, Canada, Australia, and Japan.

Approximately 1,200 total targets faced probing over a multi-week span, with successful penetrations taking around 23 hours and consuming roughly $5,000 in computing tokens. NorForge threat researchers observed the agents independently mapping secondary corporate infrastructure and exploiting control panels without human oversight. Deere, SharkNinja, and Cummins were among the corporate victims identified by independent analysis of Cairn activity logs. Commercial token economics now allow single operators to scale cyber operations at roughly $50 per successful breach.

Comment

Autonomous task execution on open-source frameworks like Cairn reduces the personnel overhead required for reconnaissance and initial access operations. Using commercial large language models to automate infrastructure mapping allows small teams to scale parallel intrusion campaigns at a fraction of traditional contractor costs.

Strategic Question for Discussion

How does the shift toward autonomous vulnerability discovery using tools like Cairn alter the cost-benefit calculus for defensive network monitoring compared to traditional signature-based detection?

The low cost of token-driven intrusion attempts points toward an operational environment where perimeter defenses face continuous, varied probes that outpace manual security patch cycles. Defenders will likely need automated remediation loops to match the persistence of agentic threat actors.

Share your assessment in the comments below.

To read in details click here