It took too long for us to appreciate the effects of social media on children’s mental health. We now face a new—and potentially more powerful—threat to pediatric mental health: artificial intelligence (AI). JAMA Pediatrics recognizes the extraordinary potential of AI to transform society, including health and health care. We have published multiple articles about it. But we are also mindful of the unique risks AI poses to children and adolescents. In particular, we are concerned about the rise of parasocial relationships with AI chatbots. Well-publicized cases of at-risk teens developing pathological attachments—sometimes culminating in suicide—are alarming and may be bellwether warnings. The Viewpoint by Radesky and colleagues lays a foundational framework for what future research should focus on. While we continue to seek the most robust study designs, we also welcome well-designed preliminary studies that clarify how substantial these risks may be, which children are most vulnerable, and what mitigation strategies are most promising. Such studies may, of necessity, rely on smaller convenience samples and proxy or aggregate outcomes—but whenever possible, they should use validated process and outcome measures and present limitations transparently. To the best of our ability, we must endeavor to make science keep pace with technology for the best interests of children.