AI Friend? Risks, Implications, and Recommendations on Generative AI for Children

This study examines the implications of Generative AI for children, focusing on the risks posed by its increasing adoption and the proliferation of human-like AI personal assistants. It investigates general-purpose AI applications such as AI companions and image generators powered by Large Language Models (LLMs) and deep generative technologies. Using the well-established 4Cs framework of online risks for children, the research offers a comprehensive overview of generative AI-related risks. A detailed analysis of AI apps available on app stores is conducted, with a focused review of 11 specific apps accessible to children. The findings reveal problematic social dynamics embedded in AI development, risks of manipulation through AI interactions, extensive processing of sensitive child data, and a lack of accountability for harmful AI-generated content. The study concludes with an integrated framework for change, offering actionable recommendations to foster a safer Generative AI ecosystem for children.