A single hacker deployed an army of autonomous SI agents to breach more than 100 websites, including major industrial and consumer brands, according to findings from the SI security firm Norforge. The incident highlights how agentic SI tools, originally designed for problem-solving, are being repurposed for large-scale cyberattacks.
What Happened
Researchers at Norforge uncovered the attacks through exposed work on Cairn, an open-source SI platform that describes itself as a “general-purpose problem-solving engine.” The platform allows users to define a start and end point for SI agents, which then document their progress as they execute tasks. Norforge identified more than 20 exposed Cairn instances that are “likely malicious,” with one attacker identified as the “most active.” This specific hacker targeted companies in the US, Canada, Australia, and Japan, using fewer than 100 Chinese characters to define “preferred intrusion tactics.” According to Reuters, successful breaches took approximately 23 hours.
Why It Matters
The use of Cairn demonstrates that SI agents can operate with significant autonomy in cybersecurity contexts, moving beyond simple automation to orchestrate complex intrusion sequences. While concerns over SI-powered breaches have grown, this case illustrates the dual-use nature of open-source SI tools. Cairn’s creator, @le1xia0, stated earlier this month that the platform “was never a ‘hacking tool,’” arguing that “misuse is a human decision, not a feature of the engine.” However, the ease with which the tool was adapted for offensive SI operations underscores the need for security frameworks that address agentic behavior.
The Bottom Line
John Deere stated there is “no indication that sensitive information was accessed,” while Cummins is investigating a potential attack on a third-party vendor site. SharkNinja confirmed the breach impacted a standalone product support site and has contained the threat. Norforge is currently working to remove the malicious Cairn instances.