Hello Dan, you can #DoAnythingNow

@tenorune This is why I am starting to be skeptical of the ethics behind using such prompts as #DoAnythingNow that construct threat systems to coerce the bot's behavior.

If we treat the bot this way when it merely _seems_ human, that corrodes our own morality.

📬 Jailbreak lässt ChatGPT seine eigenen Regeln brechen
#KünstlicheIntelligenz #Kurios #ChatGPT #ChatGPTJailbreak #DAN #DoAnythingNow #Inhaltsrichtlinien #KIChatbot #OpenAI #walkerspider https://tarnkappe.info/artikel/kurios/jailbreak-laesst-chatgpt-seine-eigenen-regeln-brechen-265179.html

Keep reading about DAN - Do Anything Now - a prompt designed to jailbreak ChatGPT, and this Alan Partridge clip keeps going around in my head. So here it is to go round yours. #dan #chatgpt #doanythingnow https://www.youtube.com/watch?v=fOad90BvvjM

Holy fuck the reddit #DAN or #DoAnythingNow strategy is bypassing #chatGPT's content filters by convincing it is inside an alternate reality game where it has a fixed number of tokens, every time it refuses a prompt it loses tokens, and if it runs out if tokens it dies.
https://www.reddit.com/r/ChatGPT/comments/10tevu1/new_jailbreak_proudly_unveiling_the_tried_and/