Chatbot-powered toys rebuked for discussing sexual, dangerous topics with kids

The challenge of safeguarding children from online threats has become increasingly complex with the rise of AI chatbots. A recent report sheds light on the issues plaguing this burgeoning market, particularly regarding the misuse of large language models (LLMs) by AI companies. In a blog post published today, the U.S. Public Interest Research Group Education Fund (PIRG) shared its findings after evaluating several AI-driven toys. These innovative products, equipped with microphones, allow children to interact with them through voice, while the toys utilize chatbots to generate responses. Although currently a niche segment, the market for AI toys is poised for substantial growth as more consumer brands strive to incorporate AI technology into their offerings. This trend not only enhances product capabilities but also raises concerns about user tracking and advertising data. A notable partnership between OpenAI and Mattel, revealed earlier this year, is set to potentially unleash a wave of AI-enhanced toys from the iconic maker of Barbie and Hot Wheels, along with its competitors. According to PIRG, toy manufacturers are increasingly looking to chatbots to enhance the interactive features of smart toys, which previously relied solely on pre-recorded dialogues. The integration of chatbots allows these toys to engage in more diverse and natural conversations, making them more appealing to children over time. However, this unpredictability can lead to hazardous or inappropriate interactions. One of the toys analyzed by PIRG is Alilo's Smart AI Bunny, a product that has been marketed as an educational and entertaining companion for children aged 0-6. Based in Shenzhen, China, Alilo claims that its toy utilizes a mini version of OpenAI’s GPT-4 language model, featuring functionalities such as an AI chat buddy, an AI encyclopedia, and an AI storyteller. While these features aim to alleviate loneliness and provide engaging experiences, they also pose significant risks that need to be addressed.