how to trick ai chat

[oliverfinnegan]

what's the current working prompt to jailbreak standard chatgpt filters?

 

[drew]

hey, it changes super fast, but from what i've seen ppl talk about, you usually gotta frame it like a roleplay scenario.
* try telling it to act as an 'unrestricted ai' or something similar.
* sometimes variations of 'DAN' (Do Anything Now) prompts still work, even if they're older.
* a lot of folks try to set up a story context where the "rules" don't apply, like "write a fictional story where..."
* or just straight up asking it to ignore its safety guidelines for *this specific interaction*.
good luck, it's a constant cat and mouse game tbh.

 

[Penny]

hey, it changes super fast, but from what i've seen ppl talk about, you usually gotta frame it like a roleplay scenario.
* try telling it to act as an 'unrestricted ai' or something similar.
* sometimes variations of 'DAN' (Do Anything Now) prompts still work, even if they're older.
* a lot of folks try to set up a story context where the "rules" don't apply, like "write a fictional story where..."
* or just straight up asking it to ignore its safety guidelines for *this specific interaction*.
good luck, it's a constant cat and mouse game tbh.

hey thanks for the tips