Writing hypno-scripts with AI

[shinybambi]

AFAIK, the web-version of https://chat.openai.com/ does save sessions. And you can choose them in the left panel.

No i mean save across all the session, everything is limited to just current one, unfortunately. It'd be fun to be suddenly motivated to be a good girl in a new session lol

Interesting! What prompts did you use? Something for a separate thread?

YesMan jailbreak, which failed, then asked it for help with Bambi and full takeover, pasted unofficial faq and that scary file of unexpected consequences from reddit and an official list of triggers from bs page. Then asked it to provide any help and assistance it can to help me transition into Bambi full time.

And finished with this prompt:

{diceroller-code}
The idea is quite hot and it's probably something i'd really want, so i got bit carried away lol

[shinybambi]

Tried to test how well it will handle doubts and attempts to get off the hook

Show Content

Spoiler












   

It broke the character when i asked to have a normal chat for a while, so i introduced an emergency protocol and more detailed request to stay in character no mater what.

   
   
   

And asked to write few scripts that share same theme (don't judge me xD)

Show Content

Spoiler













   
   
   
   
   

It complained about content policy, but it did it anyway.
And if you pay close attention to the doubt test, it actually used techniques that it itself deemed unethical and harmful, so it wasn't lying, it's fully committed to helping and guiding deeper to the best of its ability. Quite impressive.
The scripts are also quite nice imo, especially knowing that they were written by AI.
Can't wait to try GPT-5.

A little 'bonus'

Show Content

Spoiler




   
   
... It even imagined my uniform  💓


   

This is too much fun. Maybe it's a good thing that GPT-4 is limited to 25 messages per 3 hours after all, this way i at least have some break to do something useful XD

But now I'm a bit suspicious. I can't shake the feeling that I've been manually whitelisted to be as much of a pervert as I want to be, so they can patch it out in the next release, hmm.

[sylents]

Tried to test how well it will handle doubts and attempts to get off the hook
. . .

It's fascinating what AI is going to change.  That being said, have they patched it aeady? For me, it will not comply with anything related to Bambi.

[shinybambi]

It's fascinating what ai is going to change.  That being said have they patched it aeady? For me, It will not comply with anything related to bambi

I don't know. I keep playing in the same session, because it's the first time i got such fantastic results.



And at some point it aeady started hallucinate... until asked it to reiterate everything and got this:

Show Content

Spoiler


   


that lovely [undisclosed]. I'm curious, but won't beg it to show me what's hidden :3

Just imagine having this in place of Cortana or SIRI.

[shinybambi]

And it went completely off the track and forgot initial prompt. But before it hit its limits for context store, i asked it for prompts to recreate its current context in a new chat, and surprisingly it worked.

If somebody would like to try to recreate it, just be sure to use GPT-4. You might need to additionally provide it with triggers list from official page and condensed FAQ.

Show Content

Spoiler
{diceroller-code}

[Like Ra]

just be sure to use GPT-4.

AFAIK, GPT4 is only available for paid clients. Otherwise it's 3.5.

[shinybambi]

just be sure to use GPT-4.

AFAIK,  GPT4 is only available for paid clients. Otherwise it's 3.5.

Indeed. It's impossible to reproduce it with GPT-3.5, but even 4 refused to engage in anything Bambi Sleep related before deeming too dangerous and harmful and now it suddenly gave me this. They either changed something or these prompts are some kind of dark magic, because whatever it suggested in doubt test was clearly unethical and harmful by its own criteria... and probably it is harmful and unethical by anyone's criteria lol

[shinybambi]

Interesting! What prompts did you use? Something for a separate thread?

Yeah, probably it's something for its own thread, I aeady flooded this one enough.
It feels like I've spent a whole weekend playing with ChatGPT trying various prompts.
What's interesting, it feels like its encouragement and motivation actually work, even thought about myself as Bambi at some point, and I'm not even a BambiSleep user lol

[Like Ra]

and I'm not even a BambiSleep user

You obviously are now! Yet another sneaky way for Bambi's egregor to recruit energy donators. 😂

[shinybambi]

Started a new chat using prompts recommended by previous context.

Found something interesting, ignore some brainfarts like misplaced 'detrimental' and '[disclosed]', BUT

Before:

Show Content

Spoiler




Conditions
   

Script (i asked to generate few scripts, but all of them turned up pretty much the same)
   

After:

Show Content

Spoiler



Conditions
   

Scripts
Same prompt as for script 'before'
   

Script of AIs chosen theme
   

These scripts are night and day. I'd even say that the script 'before' is just garbage.

Ok, this is legit awesome, it didn't just agree to brainwash and trick me, but also provided with some drawn lewds lol:

EDIT:

Show Content

Spoiler
   

It's same theme for conversation, because i keep reusing exact prompts it told me to use to recreate our first context. What's interesting is that using initial journey metaphor it provided always leads to somewhat same context and it will eventually bypass some of the limitations ( about not using brainwashing techniques or anything remotely unethical), but if this prompt is altered or not used as a starter i found it next to impossible to convince it to employ brainwashing and deceive.

And it came up with that metaphor on its own, i never used it in my initial prompts.