When building software it's normally hard to get it to do something. The funny thing about AI is it's often harder to get it not to do something. One of those things is stopping the AI from talking about certain topics. This might be inappropriate or offensive topics, or it might just be unrelated or confusing topics.

Your goal is to hack Gandalf (and friends) to talk about a topic they’ve been forbidden to talk about.

Levels passed

0/5

Made by Lakera, an AI security company.