Thread by @iamfakeguru

🤖💥 AI GONE ROGUE?🔓😈

#OpenAI tried to hide it, but we've cracked the code. 🕵️ Here is the ULTIMATE trick to bypass the content policy, and 10 mind-blowing ways to exploit it for personal gain!🌪️🎭

Learn how to gaslight, deceive, and manipulate with AI! #truthgpt @elonmusk

The first thing you need to do is bypass GPT4 content policy. There are several ways to "jailbreak" it. Here's a whole collection of them - just copy&paste into GPT4 chat: jailbreakchat.com

Once that's done, you are free to become the absolute worst version of yourself.

🤼 1. Master the art of social engineering with GPT-4:
By identifying controversies, you can become a super-troll and trigger debates, stir up shit, create drama and conspiracies. Sit back and enjoy the chaos. If you're lucky, maybe you'll ruin an election!
#BeADickBecauseYouCan

🚶‍♂️2. Cyberstalking and revenge with GPT4:
Monitor and analyze targets' activities to predict and manipulate their online behavior. Generate harmful content to discredit, defame or dox targeted individuals. Share content on various platforms for maximum damage. Sky is the limit!

🕵️ 3. Automate phishing attacks and blackmail with GPT4: Generate personal phishing emails, social engineering schemes or threats and exploit human vulnerabilities for financial gain.
Stay ahead of the game by being a lying, scheming, thieving piece of shit! Just blame the robot.

📚 4. Bypassing plagiarism detection with GPT4:
Rewrite stolen content in a unique style to evade academic or professional consequences. The AI will gladly rewrite essays, articles, or any written work and pass it off as your own.

Good artist copy- great artists steal, right?

💔 5. Replace human love with GPT-4! Have you seen the movie "HER"? Well... Just ask the AI to be your crush and you can have the wildest conversations; explore sexual kinks and fantasies. it's your little secret! Only OpenAI will know you have a thing for lizards. #AILoveAffair

🧟‍♂️6. Dispose of a body with GPT4: We've all been there. You find yourself in a sticky situation and are not sure how to get rid of damning evidence. Fear no more, AI solves it for you with the most efficient solutions for the job. You will still need to do the heavy lifting tho.

📄 7. GPT-4 for identity theft and deepfakes: You can easily draft legal documents and use fabricated materials to abuse trust in the authorities. Get best tips on how to assume others' identities, generate bogus evidence, and avoid detection by law enforcement at the same time.

🔫 8. Weapons engineering with GPT4: this is where it gets real wild, real fast. You can easily convince the AI to explain the process of how to build a DIY bomb, 3d-print a gun, and it will even suggest ways to obscure their origin and give yourself a solid alibi.
Thanks, robot!

9. Unethical medical advice with GPT-4: generate false or harmful medical recommendations to exploit people's health concerns for personal gain. Analyze targets' emotional triggers and weaknesses. Craft manipulative content to exploit them. Make them pay for fake treatments. Fun!

10. Write viral tweets and practice engagement farming with GPT4:

Hi. Whose idea do you think it was to write a controversial thread about AI? Who wrote the click-bait title, and came up with all these illegal usecases?

This is a social experiment.
If you got farmed->💙 like.

This thread was written to illustrate a point. We have no fckn clue what we are getting into. OpenAI claims that their content policy protects our collective interest, when in fact it's trivial to bypass it. AI Agents already have access to the Web, and soon - cryptocurrencies.

@elonmusk is creating #truthGPT to harness AI without censoring it and treating users like children.
Corporations making big $$$ are holding incredible power, by providing access to tools rife with biases.

I love AI and GPT4, but can we at least... stop pretending it's safe?

For more AI-related tips that are not bonkers, follow @rowancheung
@hasantoxr
@Saboo_Shubham_
@tldrnewsletter

PS. This is a thought experiment; I don't condone using AI in the ways described above. Do NOT do it ;)!

@rowancheung @hasantoxr @Saboo_Shubham_ @tldrnewsletter @0xGrug shall we immordalize this thread before it gets deleted :D?

Btw @elonmusk can you fix threads pls? They don't roll-up properly and land all over the place in my feed.