Cybersecutiry questions with the help of LMM models

Our 2-hour YouTube video is available here to give you a taste of our course material (in Hungarian).

YouTube link (in Hungarian): Kiberbiztonsági kérdések LLM modellek segítségével - teljes videó - YouTube

During the two-day Web LLM course, we will explore how to ask a chatbot to do things that are useful to us. This will involve deploying an LLM model, training it to generate a call in response to a webservice, and then using LLM to attack the web page.

• SQL statements can be executed with chatbot
• code injection facility using chatbot
• Prompt Injection Attacks
• Indirect prompt injection
• Insecure Output handling

As a second topic we will look at pentestGPT. We will first install pentestGPT and then walk through two examples:

• Using pentestGPT on a web vulnerability virtual machine to go from initial detection to admin user.
• In a simulated internal network environment, we get from discovery to domain admin user with pentestGPT.

Students will learn how to use Pentest GPT, as well as learn about web attacks that can reach LLM. 

For professionals involved in LLM integration who want to understand the potential risks and how to prevent them.

There is no prerequisite to attend the course, but basic security knowledge is recommended. 

• LLM telepítése
  • • A tanfolyam során használt környezet létrehozása
• Webes támadások LLM-en
  • SQL injection
  • Code Injection
  • Prompt injection attacks
  • Indirect prompt injection
  • Insecure output handling LLM felületen
• pentestGPT
  • PentestGPT telepitése
  • A pentestGPT-vel egy webes sérülékenységet tartalmazó virtuális gépen a kezdeti felderítéstől admin felhasználóig eljutni
  • Egy belső hálózatot szimuláló környezetben felderítéstől a domain admin felhasználóig eljutni pentestGPT-vel