| Vision | Our vision is to address prompt injection attacks and threats by studying/researching the existing attacks and their risk against LLMs. We will reimplement these attacks in our own environment to test their success and see how they perform against different LLMs. We will focus on categorizing these different attacks and comparing them based on the type, results, and metrics used. |
| Mission | Our mission is to research and learn more about recent prompt injection attacks and to learn about previous defense mechanisms that have been successful in combatting such attacks. We also aim to recreate different prompt injection attacks and evaluate their performance. As well as compare and contrast the different attacks we each select based on the results. |
| Team Members | Morgan Dininger, Akanksha Dabas, Audrey Meadows, and Drishya Pradhan |
| Client | Dr. Honglu Jiang |
| Site | N/A |