Explore the power of language manipulation in chatbots and large language models in this 41-minute conference talk from the 37th Chaos Communication Congress (37C3). Discover how creative wordplay and social engineering techniques can be used to make AI systems serve human purposes, often in ways unintended by their creators. Learn practical strategies for leveraging language skills to overcome the limitations and biases of chatbots, potentially uncovering hidden information and assisting in investigative research. Gain insights into protecting personal data while understanding how to ethically exploit the vulnerabilities of AI language models for positive outcomes.