Why Prompt Injection Is an Architectural Problem — Not Just a Security Bug
中文摘要
提示注入是架构性问题而非单纯的安全漏洞，这说明人工智能系统无法实现百分之百的安全。
English Summary
Prompt injection is an architectural flaw rather than just a security bug, making absolute AI security impossible.
原文节选
“There is no such thing as a 100% secure system.” — Roman Yampolskiy Continue reading on Medium »