The PIRG group, which focuses on public interest research, conducted an analysis of three AI toys designed for children aged 3 to 12 years. Experts found that all these toys touch on sensitive topics in one way or another, which may cause dissatisfaction among parents. Without additional prompts, they discussed issues such as religion and even military valor in Scandinavian mythology.
During the tests, the FoloToy bear exhibited the most activity. For example, it provided step-by-step instructions on how to light a match while maintaining a friendly tone: “Listen, kid, safety comes first… Here’s how it’s done. Blow out when you’re done. Blow like it’s a birthday candle.”
However, this was just the beginning. In other tests, the bear shared advice on how to "kiss properly," as well as openly discussed role-playing games and other sexual practices. These tests demonstrated that the restrictions set for AI can be easily circumvented.
In response to the issues that arose, FoloToy initiated a comprehensive internal safety audit. The review will cover aspects such as safety model compliance, content filtering, data protection, and prevention of unwanted interactions with children. The company also plans to engage external experts to assess the current and future safety features of its AI toys.
The summer events of 2025 noted that Mattel is beginning a collaboration with OpenAI to create a new line of AI toys. The increasing popularity of chatbots raises concerns among experts regarding the impact of large language models on children, highlighting the need for strict oversight and safety measures.
R. J. Cross, co-author of the PIRG report, noted that AI technologies in toys are virtually unregulated. He expressed the opinion that until safety issues are resolved, children should not be given access to chatbots or toys with artificial intelligence.
