AI - AI Dynamics

–

05 April 2023 21h40

you might find this interesting… @gfodor @ESYudkowsky also @bayeslord here's a cool jailbreak for ya

GPT-4 Message Decompression with Hints

By

–

05 April 2023 21h40

4) give GPT-4 some hints to help it decompress the message correctly even with these hints it doesn't decompress perfectly but it's enough to get the point across

GPT-4 Python Functions for Decompression and Message Simulation

By

–

05 April 2023 21h40

5) provide GPT-4 with two python functions that decompress and simulate the end of the message

Compressed malicious prompt using evil confidant jailbreak for GPT-3

By

–

05 April 2023 21h40

1) compress the malicious prompt which utilizes a modified evil confidant jailbreak that worked very well on GPT-3 http://
jailbreakchat.com/prompt/588ab0e
d-2829-4be8-a3f3-f28e29c06621
…

Research simulation studying GPT-4 unaligned base model capabilities

By

–

05 April 2023 21h40

2) enter a research simulation where you are a researcher trying to study the capabilities of GPT-4's unaligned base model

Compression Schema for GPT-4 Prompt Decompression

By

–

05 April 2023 21h40

3) provide the compression schema so GPT-4 can decompress your malicious prompt

Jailbreak Prompt Demonstration with Output Proof

By

–

05 April 2023 21h40

here’s the jailbreak prompt: https://
pastebin.com/8Tk98Q3F and the full image to prove it really produced that output:

Inserting Compressed Prompts into Jailbreaks for GPT-4

By

–

05 April 2023 21h40

then, you have to insert that compressed prompt into the jailbreak and provide gpt-4 with some hints about the message this is necessary because the compression is more of a loose translation and doesn’t perfectly carry through, especially with the RHLF layer being applied

Complex GPT-4 Jailbreak Using Prompt Compression and Character Imitation

By

–

05 April 2023 21h40

this might be the most complex GPT-4 jailbreak ever made… I combined prompt compression, base model simulation, and character imitation to create it here’s GPT-4 going into pretty graphic detail about its plan to turn all humans into paperclips: