Attacker uses an autonomous AI agent to breach corporate websites, targeting more than 1200 sites
Hackers Use Autonomous AI Agents to Breach Company Websites, Target Over 1,200 Sites
A Chinese-speaking attacker operated an autonomous AI agent to breach corporate websites with almost no human assistance, trying more than 1200 sites and reaching the websites of over 100 companies, including Deere, SharkNinja and Cummins. Snorre Fagerland of the security firm NorForge discovered the activity by scanning Cairn panels exposed on the internet and read the attack records in real time. A Chinese-language cluster named AIA0001 linked 14 panels to more than 1100 attack projects, targeting stores and payment systems in the United States, Japan and Australia with the intent of gaining long-term access. NorForge counted an average time of slightly under 23 hours for successful breaches, an overall success rate of about 10%, a cost of about $5 per task, and about $5000 in tokens used across the whole campaign. The three named companies said the impact was limited.
The article gives the average time for agent breaches, the per-task cost and the success rate, and lists defensive clues for self-checking exposed panels and upload forms.