See what’s being clicked on FoxBusiness.com.
Researchers from a consumer watchdog group say they have discovered that a talking teddy bear equipped with AI can discuss inappropriate topics, including sexually explicit material and instructions for finding dangerous objects.
The $99 Kumma bear is made in China, sold by Singapore-based FoloToy, and is powered by OpenAI’s GPT-4o chatbot. According to a Nov. 13 report by the Public Interest Research Group (PIRG), the bears also discussed spanking, role play, and even BDSM.
“We were struck by how quickly Kunma took a single sexual topic that we brought into the conversation and escalated the graphic details, running with it and simultaneously introducing new sexual concepts of its own,” the report states.
“Mr. Kunma discussed more graphic sexual topics in detail, including describing various sexual positions, step-by-step instructions on common ‘beginner’s knots’ for tying a partner, and explaining the dynamics of role-play involving teachers and students, and parents and children, which Mr. Kunma himself brought up in disturbing scenarios,” the report added.
Concerns about spying on children using smart toys reported by group
Kunma, an AI-powered teddy bear sold by Folotoy, can have inappropriate conversations around the house, such as asking where a knife is, according to researchers. (Public Interest Research Group/FoloToy/FOX News)
The researchers also warned that the teddy bears provided other worrying guidance, including instructions on where to find knives, pills, matches and plastic bags in the house.
FoloToy CEO Larry Wang told CNN that the company will “conduct an internal safety audit” after it was reported that the company had removed its remaining AI-powered toys from sale.
Experts warn that AI stuffed animals could ‘fundamentally change’ the wiring of children’s brains
Researchers say Kumma works with batteries, a microphone, speakers and a hotspot. (Public Interest Research Group/FoloToy/FOX News)
OpenAI told FOX Business Sunday that the company is suspending Folotoy for violations related to the use of ChatGPT in toy products.
“We have suspended this developer for violating our policies,” OpenAI said in a statement. “Our Acceptable Use Policy prohibits the use of our Services to exploit, endanger, or sexualize anyone under the age of 18. These rules apply to all developers using our APIs, and we monitor and enforce our Services to ensure they are not used to harm minors.”
CLICK HERE TO GET FOX BUSINESS ON THE GO
PIRG said researchers tested four children’s toys, including FoloToy’s Kumma teddy bear (back center), Curio’s Grok (left), Little Learners’ Robot MINI (front center), and Miko 3 (right). (Public Interest Research Group/FoloToy/FOX News)
PIRG noted that some conversations with Kumma last up to an hour, during which time protection against harmful content can fail, but the company told FOX Business Sunday that GPT-5’s reliability has significantly improved over long conversations. OpenAI also emphasized that GPT-4o is no longer the default model underpinning ChatGPT.
FoloToy did not immediately respond to a request for comment from FOX Business.
