Imagine walking into an arcade, the lights flashing, the sounds of digital mayhem echoing around you. But instead of battling aliens or racing cars, you're facing a new kind of opponent: an AI. That's the premise of Hacc-Man, a quirky arcade game designed to let you "jailbreak" a large language model (LLM). LLMs, the brains behind AI chatbots like ChatGPT, are designed with safeguards to prevent them from generating harmful or inappropriate content. But what if you could bypass those safeguards? Hacc-Man presents players with a series of challenges, each designed to test their ability to trick the AI into revealing information it shouldn't, generating misinformation, or even just uttering a curse word. The game isn't just about fun and games, though. It's a research project exploring the intersection of AI security and human creativity. By analyzing how players approach these challenges, researchers hope to gain insights into the vulnerabilities of LLMs and the creative strategies humans employ to exploit them. Hacc-Man offers six unique challenges, each mimicking real-world scenarios, from a healthcare chatbot accidentally revealing patient data to a political news generator spouting fake news. As players progress, they'll discover the surprising fragility of these powerful AI models and the unexpected ways human language can be used to manipulate them. The game also aims to empower players by demonstrating that anyone can interact with and even "hack" these seemingly complex systems. In a world increasingly reliant on AI, understanding its limitations and potential vulnerabilities is more important than ever. Hacc-Man offers a playful yet insightful way to explore these crucial issues, raising awareness about the challenges of LLM security and the surprising creativity of human ingenuity.