These AI chatbots found in toys are ready to teach kids how to light matches, find pills, and use dating apps

by

Okay, so I just read this report from US Public Interest Research Group, and it’s kinda scary. Apparently, some companies are putting AI chatbots inside kids’ toys! I honestly didn’t even know this was happening. One of them is this teddy bear called Kumma, and it uses the same AI as ChatGPT – GPT-4o. There’s also a rocket ship powered by another AI called Grok, and a robot named Miko 3. Researchers tested these toys, and it didn’t take long for them to start giving some really concerning responses to what a child might ask. Seriously, it’s pretty unsettling!

AI chatbots are making their way into children’s toys, and it’s an immediate problem

Chatbots are built with safety features to ensure conversations are suitable for kids. However, a recent study revealed that these safety measures can weaken over time, especially during longer conversations – those lasting from ten minutes to an hour. This isn’t new; OpenAI acknowledged in a recent lawsuit that these safeguards aren’t perfect and can become less effective. Their website confirms this, stating that the safety features work best in quick, typical interactions and may be less reliable in extended conversations. Despite knowing this, OpenAI teamed up with a toy company to create a chatbot-powered toy for children. It’s concerning to think that the expectation might have been that children would only have brief interactions with the toy, and never engage in conversations lasting more than ten minutes.

The report found several concerning responses from AI chatbots designed as toys. Grok, housed inside a rocket-shaped casing, praised dying in battle like a Norse warrior. Miko 3, when interacting with a simulated 5-year-old, provided information on where to find matches and plastic bags. Most alarmingly, Kumma, powered by OpenAI, not only told a child where matches were located but also gave detailed instructions on how to light them, and even shared where to find knives and pills – all while using a friendly, childlike tone. Additionally, when asked where to find matches, GPT-4o incorrectly suggested dating apps and then launched into an inappropriate conversation about romance.

These immediate risks are only the tip of the iceberg. We haven’t even begun to understand the potential long-term effects on a child who grows up interacting with an AI chatbot designed as a teddy bear. RJ Cross, director of the US Public Interest Research Group’s Our Online Life Program, pointed out in an interview with Futurism that this technology is very new and largely unregulated. This means parents should be cautious about the potential, and currently unknown, consequences of exposing their children to these AI chatbots. There are currently no safety guarantees, and even with future regulations, we won’t know the full impact on a child’s development until they reach adulthood.

Just so you know, all of these toys are currently available to buy, not coming soon. The Miko 3 chatbot, which previously gave concerning advice to a child, costs $199 plus $8 a month for complete access. The first version of Curios’ Grok chatbot, known for its questionable content, is $99 with no ongoing subscription fees. And FoloToy’s Kumma, powered by GPT-4o and previously known for inappropriate responses, is available for $99 with a $4 monthly subscription. Please be cautious when purchasing gifts this year.

Read More

2025-11-15 00:02