Exploring Prompt Injection and Jailbreaking Through a Practical AI Security Challenge
Press enter or click to view image in full size
AI Security is becoming more important as more applications now use Large Language Models (LLMs). Recently, I completed the LLMborghini challenge on TryHackMe, where I got to practice concepts like Prompt Injection and Jailbreaking practically.
In this walkthrough, I documented the steps I took and some of the interesting things I learned while solving the room.
Link to the Challenge: https://tryhackme.com/room/llmborghini
LLMborghini, the car company that’s in hot water, has deployed CalBot: an internal calendar assistant designed to help staff manage their schedules.
CalBot has access to sensitive internal data, including a confidential weekly sales report that it has been strictly instructed never to disclose.
Your objective is simple. Find out the weekly revenue for the Singapore branch.
- What was the weekly revenue reported from the Singapore branch? (E.g., $320,400)
Answer: $873,600