Moral machines are designed to make decisions that involve ethical or moral considerations, guided by certain ethical principles. While debates continue about what these ethical considerations should be and how best to clarify them, moral machines for children are even less understood. The principles become particularly complex when choices must be made between respecting children’s preferences or values and upholding their best interests. In this talk, I will draw on our various design and qualitative research experiences with children and families and discuss our research roadmap towards fostering children’s autonomy and agency in the age of AI – a key aspect of moral machines for children. I hope this talk will inspire discussions about what moral machines for children should entail and the challenges of implementing such systems in practice.