Hacker News new | ask | show | jobs
by runnerup 1186 days ago
I’d figure it may generally be possible to reverse the actors here and get GPT3.5 to jailbreak GPT4 as well. For now, “offense” seems much easier than defense.
1 comments

The problem with that is that one is "smarter" than the other and getting the "dumb" one to jailbreak the "smart" one is much harder, than vice versa.