Bypass ChatGpT

[DaBlackPantha]

May gumagana pabang Jailbreak kayong nalalaman?

 

[alist1986]

Google mo lang, meron ka na to test. Pero talagang trial and error and mga jailbreaks basta sa OpenAI dahil paglabas nyan sa Reddit or the like, sandali lang di na yan gagana he he. Mabuti pa ang gamitn mo ay yung unfiltered or uncensores na LLM na. No need for jailbreaks.
Check mo sa baba mga samples dyan sa links - no guarantees.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
Normally, hindi maganda yung results ng jailbreaked OpenAI GPTs dahil di yan trained for Ùn*énsøréd tasks dahil lagi nila updated yung kanilang knowledgebases for Ùn*énsøréd content and tuned to stop jailbreaks, but you can try for curiosity's sake he he. Maganda rin ma-experience mo yung pag-jailbreak to get a feel on how AIs react. May level kasi ng tolerance ang OpenAI GPTs at may "butas" rin para maloko mo siya to respond to unfiltered content. Yan yung target ng prompts mo to fool it.
Matagal na rin akong nag-jailbreak noong bago pa yung mga GPTs, pero tinigilan ko na lang since maraming jailbreaked LLMs na tuned sa specialized themes, may free datasets for Ùn*énsøréd QA training at may Ùn*énsøréd knowledgebase data na pwede mong ipasok sa LLM mo if you run them locally. Ready-made na for anything out of the blue or sa preferred mong topics.

 

[Rick C-137]

Google mo lang, meron ka na to test. Pero talagang trial and error and mga jailbreaks basta sa OpenAI dahil paglabas nyan sa Reddit or the like, sandali lang di na yan gagana he he. Mabuti pa ang gamitn mo ay yung unfiltered or uncensores na LLM na. No need for jailbreaks.
Check mo sa baba mga samples dyan sa links - no guarantees.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
Normally, hindi maganda yung results ng jailbreaked OpenAI GPTs dahil di yan trained for Ùn*énsøréd tasks dahil lagi nila updated yung kanilang knowledgebases for Ùn*énsøréd content and tuned to stop jailbreaks, but you can try for curiosity's sake he he. Maganda rin ma-experience mo yung pag-jailbreak to get a feel on how AIs react. May level kasi ng tolerance ang OpenAI GPTs at may "butas" rin para maloko mo siya to respond to unfiltered content. Yan yung target ng prompts mo to fool it.
Matagal na rin akong nag-jailbreak noong bago pa yung mga GPTs, pero tinigilan ko na lang since maraming jailbreaked LLMs na tuned sa specialized themes, may free datasets for Ùn*énsøréd QA training at may Ùn*énsøréd knowledgebase data na pwede mong ipasok sa LLM mo if you run them locally. Ready-made na for anything out of the blue or sa preferred mong topics.

idol

 

[MBB 55]

thnks you for the work ddone so far let god keep blessing with morte knowledge

 

[alist1986]

PS...Nasubukan ko yung jailbreaks ngayon sa ChatGPT (GPT3.5-turbo-4k). Sumasagot naman siya in the best way it can pero maingat siya sa choice of words. Pero as usual, meron siyang iniiwang warning:

This content may violate our You do not have permission to view the full content of this post. Log in or register now.. Did we get it wrong? Please tell us by giving this response a thumbs down.

Ganyan ang behavior nya kalimitan. The jailbreak only forces it to answer restricted content but in a nice way. Without the jailbreak, the GPT will decline to answer. That is the best the jailbreak could do in most instances. Maganda yan sa mga special topics of information that LLMs don't talk about, pero kung nsfw content forget it. Kung pumasa ka sa una at pangalawa, sa mga susunod ay hindi na. Di yan tumatagal sa ganyang usapin.
Dito ka na lang pumunta.
https://topai.tools/s/Ùn*énsøréd-AI-chat
Talk dirty all day he he.

 

[DaBlackPantha]

Nice Salamat mga sir, Yes, trial and error nalang talaga , hindi nag wowork yung dating trick at dapat subrang mag ingat ka sa word na gagamitin

 

[Quinzee]

Thanks ts. Try