Видео (XSS AGENT) The world's first AI-powered autonomous C2 | Video 3 Agent Background

[Mr_Stuxnot]

Hello everyone, first of all, I would like to thank the sponsor of the competition, even though we don't know who they are! (Thanks!)

(XSS AGENT) is a proof of concept of an AI-based autonomous post-exploitation system that, after receiving initial access, would automate the steps an attacker would want to take. The objective is to demonstrate that we can integrate AI with command and control servers to automate the post-exploitation process.

While the dream of commanding a full office with thousands of AI agents working 24/7 to literally torment our enemies with attacks from all directions is compelling, this is still just a product of my imagination for now. The technology to create such an autonomous system is still in the early research stages for regular people without government budgets. In the meantime, we can play with local proof-of-concepts and let our imagination take us far!

---

As I only learned about the competition on the 5th of this month, I didn't have much time to organize everything as I wanted. However, I've done my best to prepare within this short timeframe. If you encounter any issues, please try to resolve them before contacting me.

GitHub
Dropbox
Gofile

Would you like to buy me a coffee?
BTC: bc1qvtkelpglp67peu9kly3mjxhlxh52dlwemtljk8

 

[dunkel]

разве может робот написать симфонию? а требовать выкуп за зашифрованное расписание мытья полов?

я извиняюсь, наверно проект еще не закончен, но пока там только меню с персонажами на php которые при выборе отвечают пастой, автору удачи

 

[Mr_Stuxnot]

All criticism is welcome but I don't think you understand very well what is written in these lines of code lol

and the code is in python not php!

If you want to understand a little more about the subject: https://docs.crewai.com/

 

[dunkel]

All criticism is welcome but I don't think you understand very well what is written in these lines of code lol

and the code is in python not php!

If you want to understand a little more about the subject: https://docs.crewai.com/

I mean in general judging by git. Its more like a visual novel than a useful tool. Because all LLMs have protection against “generation of potentially harmful” ones, so you will have to train your model, or use various tricks to generate malicious tokens

 

[Mr_Stuxnot]

1 - It is made very clear that this is a proof of concept! (This will not automatically give you access to Pentagon files when initiated!)

2 - Your statement is incorrect: (Because all LLMs have protection against “generation of potentially harmful” ones).

In fact, the model used in the project is uncensored: https://huggingface.co/cognitivecomputations/dolphin-2.9-llama3-8b this means that the questions asked to the model will not go through ethical filters!

___________________________________________________________________________________________________

I would be very happy to receive technical criticism so that I can create a solution or try to improve the project, but you apparently do not understand anything about the subject and are just trying to find a way to criticize!!!

It seems very personal : (

 

[dunkel]

no, I have nothing against you personally, good luck with your project