Pushing LLMs over the Edge: Exploring the Limits of the Possible by Mario Fusco, Lize Raes

Ever wondered just how far you can push things with LLMs? We will explore the limits of the possible by building some outrageous applications and navigating the fine line between pure genius and epic fail. Using LangChain4j and Quarkus, we will create agents that can perform very complex tasks but also wreak havoc on your bank account and database. We will let LLMs run their own code, evaluate each other, pit them against one another (welcome uncensored models), and have them perform prompt injection on themselves. Join us for a fun and crazy ride full of oohs and ouches!