News for hackers. Stuff that matters.
Re:ChatGPT Hacks by thomasahle on Tuesday July 25, 2023 @01:42AM (#36858411)
It's interesting that many ChatGPT "prompt hacks" also work on other language models. I tried copying the DAN (Do Anything Now) prompt into the LLaMa 2 system prompt, and it worked more or less unchanged: https://twitter.com/thomasahle/status/1683508252821291008

There are no comments on this comment yet.