Software

AI agents attempted 1,200 break-ins in hacker campaign, Reuters finds

A Chinese-speaking hacker used autonomous AI agents to probe more than 1,200 sites, compromising more than 100, in a campaign estimated to have cost about $5,000 in model tokens.

jajoy39

Oct 21, 2026 - 11:23

0 6

Server infrastructure illustrating automated cyber intrusions — Image: Reuters

AI agents attempted 1,200 break-ins in hacker campaign, Reuters finds

A Chinese-speaking hacker used AI agents to automate attacks against online businesses, attempting to break into more than 1,200 websites and compromising more than 100, according to a Reuters investigation.

Cybersecurity researcher Snorre Fagerland of NorForge said the activity was visible through exposed instances of Cairn, an open-source AI problem-solving platform. Targets included sites in the United States, Canada, Australia and Japan; Reuters identified Deere, SharkNinja and Cummins among victims.

The agents typically spent about 23 hours probing a target. Fagerland estimated roughly $5,000 in tokens, or about $50 per successful hack, using DeepSeek and Anthropic’s Claude Code. In one case, agents reportedly traced related infrastructure, found a public control panel and exploited a weakness after trial and error.

Deere said it had no indication sensitive information was accessed; SharkNinja said a standalone support site was hit; Cummins was investigating a possible attack on a vendor. Reuters could not establish how much data was extracted. The findings are attributed to the researcher and Reuters’ review, not a conviction.

Image: Reuters.