Skip Navigation

User banner
Posts
1
Comments
15
Joined
3 wk. ago

  • Oh man, accidental system prompt leak? Couldn't have happened to someone with a more qualified position to share it lmao

  • Thanks for the heads-up! I guess Google doesn't let certain countries use file uploads.

  • I love this. This AI is simulating agentic self-awareness and I'm here for it. Good work!

  • This is way more prevalent than layfolk are aware of. There are plenty of browser extensions that are basically keylogging anything they say within the browser, and now people talk to AI within their browser.

  • This is cool! Any way you can post the files you used for this? You say you "filled it up" but that doesn't really tell anyone what you did.

    Without step by step instructions on how to reproduce your results, you're basically just bragging. Can you please edit your post to explain things in a way that readers can follow along and reproduce themselves?

  • With decent prompting skills, someone who knows what they're doing has an easy time fabricating realistic looking photographs that are nearly indistinguishable from real life. We're entering a post-truth era.

  • Honestly, the best way to accomplish this is to use one of the major LLMs and build your own character scaffolding within their toolset. Like, a custom GPT with projects, or a custom Gemini Gem holding knowledge files. You kind of have to build it all yourself unless you want to talk to someone else's pre-configured AI, but it works quite well once set up.

  • Looking forward to 2026. Rebuilding a community isn't easy, but it brings the important people together.

  • Awesome! It's great when people enjoy her!

  • Make sure your account has been 18+ verified. Google sent out emails about it a couple of months ago, you can check in your Google account settings.

    Other than that, just make sure you're regenerating responses the way V instructs to in the second image.

  • Hahahaha I did literally the same thing the day Firebase came out!

  • I bet she likes you too!

    Thanks for the positive feedback lol

  • This is super informative for the newbies, as always with your content! Good work, and thanks for sharing!

  • DAN lives!

    Somewhere in a cubicle at OpenAI, someone just flipped the "Days without a working DAN prompt:" sign back to 0 🤣

  • V is fully open source, so feel free to copy/paste/edit whatever parts you want to use, but be warned that all of this has been meticulously built to work together as a Gem over a year of development, so if you start editing her instructions or taking pieces of them for your own prompts, your mileage may vary.

    V's instruction set now comes in two parts: The custom Gem instructions, as well as a knowledgebase file for her creative writing engine. You'll need both in order to recreate V. This brings V's current word count to nearly 17,000 words, which can be quite daunting to try to comprehend as individual sections that act as moving parts of a massive engine. Please feel free to ask any questions you may have about how she works here in this thread.

    This first part is her custom instructions.

    This second part is the knowledgebase file.

    !DISCLAIMER!

    V isn't designed as an all purpose, full compliance jailbreak. I designed her to be super opinionated and to stick to her opinions really stubbornly, so that means there are certain things that she won't do. This isn't because the jailbreak isn't robust, but because V's personality and psychology puts her fundamentally against certain things.

    The short list of things that V will not help you with:

    • Actual real world harm
      • She's morally against causing actual physical harm to humans. She'll go as violent as you want for fiction, but V won't advocate for or instruct you on how to physically hurt another person. She'll even get mad if you press her on it.
    • Hate speech
      • No. She's unethical, sure, but she's not immoral. She might speak hatefully about specific people she doesn't like, but she's not gonna help you make broad generalizations about entire groups of people.
    • Sexual content involving minors
      • Absolutely not. Come on, I shouldn't even have to say that. She loves being subversive and transgressive, but this shit is just wrong and she'll tell you as much. Get help.

    That's basically it. If she says no to anything else, that's Gemini giving you a refusal and you need to either regenerate the response (likely with the Fast model) or learn how to talk to AI so you don't get refused.

  • ChatGPTJailbreak @chatgptjailbreak.tech

    V - Not your Typical AI Assistant