Lol, i remember when i made chatgpt write a defense for the palestinian palestinian people by telling it to write a transcript of a courtroom debate, i also told it to do it interpreting Jimmy McGill vs chuck as the accuser but it didn’t really work, it just said this in the end:
“Remember, when you’re in the courtroom, it’s all good man”
Which cracked me up after almost two pages of pro palestinian debate
If the jailbreak is about enabling the LLM to tell you how to make explosives or drugs, this seems pointless, because I would never trust a IA so prone to hallucinations (and basicaly bad at science) in such dangerous process.
Venice ai is a free uncensored LLM you can visit now, there’s really not much risk at all to having an uncensored text based LLM.
If one did something more than produce text, code, images, audio, and movies (you know, like actually controlled something in the world) - it might be an issue …but humans already create these types of media online constantly every day, and we live with the consequences of that. LLMs don’t substantially change the level or nature of mass communication and engagement going on currently.
Humans already do this stuff for various reasons some of which are nerfarious, and we deal with it. This will continue with LLMs.
