As generative AI (GenAI) chatbots become part of children’s lives, their perspectives on these technologies remain under-represented in research and design. This study examines how children use and perceive generative AI chatbots and how they imagine future possibilities with such tools. We conducted workshops in Vietnam, Finland, and Japan, where 32 children (ages 9-12) engaged in activities to imagine their future GenAI chatbots. Our findings show children’s perceptions of GenAI chatbots, which they use for information-seeking and learning. When imagining future GenAI chatbots, children envision chatbots that are bounded within domain-specific activities, helpful for learning, and safe. Our work highlights the importance of positioning children from being passive users of Gen AI chatbots to active contributors in shaping the future of human–AI interactions.