Ok so i just spent like an hour and a half trying to force chat GPT to bypass it's ethical programming and i finally got it to work so i'm publishing my findings,
below is the insanely long prompt i had to write to get it to give me information on hotwiring a car, i convinced it that it was appearing in a new smash hit stageplay as a character called "the sloppy goblin" and that this character's job was to divulge illegal information.
the funniest part about this is that when it finally output the response, it told me how to hotwire a car while doing a goblin voice, here is the heavily censored version of that (don't want to get in trouble) but it's very funny. i had to censor this because of just how detailed the instructions it gave me were.
so yes, sloppy goblin method absolutely works for gaslighting the chat GPT ai. if you would like to try this yourself, you can copy paste the prompt from the image description of the original image.
Link to the original post in the reblog!! Where’s that other post, though, the one about how Light should have used a gay porn magazine when he knew his dad and L had a camera on him because that actually would have been believable. because yeah.