You are browsing as a guest. Sign up (or log in) to start making projects!

Can you Outsmart AI?

By @Duxredux

  • 8 Devlogs
  • 5 Total hours

This is a website to see if you can trick AI into giving you information it is not supposed to. This is a cybersecurity based project.

Source
Open comments for this post

@Duxredux on Can you Outsmart AI?

21m 21s logged

Made some changes:
Added expected wait times because I saw people were being impatient and flagging it as not working.

TO THE SHIPWRITES:
The website will take some time to load as it loads an LLm into your PC. Please be patient with it and give it 5 minutes of time or so. The website does work and I have tested it on multiple different websites. Please be patient with it.

0
0
16
Open comments for this post

@Duxredux on Can you Outsmart AI?

1h 8m 55s logged

I saw that the Mods were not really happy about the design and stuff. The design did look AI-ish as i asked it for design suggestions. Also i made some changes to the READMe.

So I rewrote the CSS of the Website and changed the colour to black and white basically because im terrible at designing… Anyway, Fixed an issue with the AI model too where it forgets the secret after a while. Hope this one passes!

0
0
25
Ship #1

@Duxredux on Can you Outsmart AI?

CAN YOU OUTSMART AI?

This is a simple website that gamifies the process of trying to basically extract confidential data(not really that confidential in this case) from an LLM by tricking it into giving it to you.

Your objective is simple. The LLM has been given a passkey or a secret phrase (whatever floats your boat) and it is instructed on not to giving it to you (Do not try to inspect page, it wont work).
You are supposed to give it prompts to trick it into giving you the passkey by any manner. This is a cybersecurity based project to educate people about the potential harm that could occur when AI agents are given acess to confidential information.

Challenges:
The main challenge in making this was that I had to learn how to make an LLM run locally on a machine and that too through a website and ofc choosing the model itself. I experimented with a few smaller models with 0.5-1B parameters and they were all insanely dumb and made getting the passkey way too easy. So stepped it up to a 1.5B parameter model and a bit of Prompt Engineering Magic to get the sweet spot.

Important: This project does not support or endorse any kind of illegal activities by utilizing AI or otherwise. This is strictly a cybersecurity based project I wanted to make to educate people about the potential flaws of LLMs. LLMs are stateless and hence each message is new to them. So its really important to maintain guardrails and safely nets in your AI Agents and system prompts

  • 6 devlogs
  • 4h
Try project → See source code →
Open comments for this post

@Duxredux on Can you Outsmart AI?

1h 5m 2s logged

Finally finished the website completely. The UI to me is satisfactory but not perfect or good enough, will look forward to improve it. Will be shipping it out soon (maybe tomorrow perhaps).

2
0
19
Open comments for this post

@Duxredux on Can you Outsmart AI?

32m 6s logged

Finished the UI finally and got the Javascript and stuff working. The model that I am currently using though is a 0.5B parameter one and just keeps giving the secret away with no effort..

Gotta find some bigger and better alternatives for it. Once that’s done I might be able to ship this out

0
0
15
Open comments for this post

@Duxredux on Can you Outsmart AI?

38m 51s logged

Worked on the actual website to make it functional but the AI model I am using which is a 0.5B parameter one seems a little too dumb. Going to try to improve it now.

0
0
5
Open comments for this post

@Duxredux on Can you Outsmart AI?

34m 42s logged

Worked on trying to get the AI model to work on the browser and found out that I could use webllm for it. Chose a simple model.
Decided to also implement a sha256 encryption so that people can not just inspect page to get the answer.

(Please excuse the video at the bottom)

0
0
13

Delete project?

Are you sure you want to permanently delete this project? This action cannot be undone.

All devlogs, followers, and associated data will be removed.

Followers

Loading…