Security Perimeter Stops at the AI Layer. That’s a Problem.
中文摘要
现有的安全边界未能覆盖 AI 层，这意味着即使 AI 运行正常，仍需要额外的安全约束。
English Summary
Current security perimeters fail to cover the AI layer, necessitating additional safeguards and controls even when AI operates as designed.
原文节选
Your AI is operating as designed, but it still needs a leash. Continue reading on Medium »