返回首页
AI on Medium··行业媒体

I Broke All 8 Levels of Gandalf’s LLM Defense. Here’s Every Technique I Used.

中文摘要

分享破解Lakera Gandalf大模型防御挑战所有级别的技巧

English Summary

A researcher successfully bypassed all eight levels of Lakera’s Gandalf AI security challenge, documenting the specific prompt injection techniques used to compromise the LLM’s defenses.

原文节选

Gandalf is a prompt injection challenge by Lakera. Continue reading on Medium »