This article introduces the red team exercise held at the 31st annual DEF CON, where thousands of hackers will attack the world's top generative AI models. This is a large-scale event aimed at identifying the weaknesses of AI models and providing feedback on embedded harms. The event is provided with models by multiple AI companies and is carried out in cooperation with organizations such as Humane Intelligence, SeedAI, and AI Village. The article also mentions the goals of the event and the backgrounds of the participants.