I Think LLMs Are Hackable And the Attack Surface Isn’t Where We Think
中文摘要
作者认为大语言模型（LLM）具有被攻击的风险，且其攻击面并非人们通常认为的位置。
English Summary
The author argues that LLMs are hackable and that their attack surface is not where we typically expect.
Original Excerpt
Guys, I Think LLMs Are Hackable Continue reading on Medium »