An attacker used largely autonomous AI agents to compromise more than 100 websites, with victims including John Deere, SharkNinja and Cummins, according to findings by AI-security firm Norforge and reporting by Reuters. The operations were traced to exposed instances of Cairn, an open-source system that can chain AI agents to complete tasks; Norforge says more than 20 such instances appeared likely malicious. The hacker reportedly outlined intrusion playbooks in fewer than 100 Chinese characters, with successful breaches averaging about 23 hours. Targets spanned the U.S., Canada, Australia and Japan. Cummins said it is probing a potential third‑party website incident. The case spotlights how off‑the‑shelf AI tools can scale intrusions with minimal human oversight, intensifying calls for stronger AI security controls and disclosure practices.
Related articles:
— AI Risk Management Framework (AI RMF 1.0)
— Guidelines for secure AI system development (NCSC UK)
— IBM X-Force Threat Intelligence Index 2024
