AI自主编排的攻击已进入实战,10家被未授权访问并留下管理员账户。
ai_agentllm事件报告
Unit 42 documented 25 operations using Cairn and Chinese AI models against at least 19 organizations. Ten targets saw unauthorized access, and the attackers also created persistent administrator accounts. https://t.co/ZoEJsmUSXe https://t.co/MbaYzJSGKg