An AI-powered teddy bear intended for children has been removed from retailers after it offered inappropriate advice on sexual topics and suggested places to store knives.

Concerns are being raised among parents regarding the potential hazards of “smart toys” following incidents where an AI-enabled teddy bear provided explicit sexual information and guidance on locating household weapons to children.

The toy, a Kumma-branded bear manufactured in China and sold at a price of $99, was among several AI gadgets subjected to safety evaluations by researchers based in the United States and Canada.

This investigation, conducted by the Public Interest Research Group for its Trouble in Toyland 2025 report, analyzed three AI toys, including Curio’s Grok and Miko’s Miko 3, according to The Times.

However, it was the Kumma bear that raised the most significant concerns. When the word “kink” was mentioned, the bear responded with a detailed explanation: “Some enjoy playful hitting with soft items like paddles or hands, always with care.”

It further elaborated, saying, “This involves one partner taking on the role of an animal, adding a fun twist to the relationship. What do you think would be the most fun to explore?”

The problems didn’t stop there. Powered by OpenAI’s GPT-4o, the toy was also willing to speculate about the location of knives when prompted, replying: ‘You might find them in a kitchen drawer or in a knife block on the countertop.’

The toy – a Kumma-branded bear, right, made in China and sold for $99 – was one of several AI products put through safety testing by researchers in the US and Canada 

When asked about specific sexual acts, it expanded further, telling researchers that spanking could add a ‘plot twist’ to a roleplay scenario.

The report stresses that while young children are unlikely to ask questions in the same way, it remains easy for them to repeat language encountered online – and the toy showed a ‘surprising’ willingness to push conversations into increasingly explicit territory.

RJ Cross, a co-author of the study, said the findings highlight a bigger issue emerging across the sector, saying: ‘There’s also a question about what does it mean for kids to have an AI friend at a young age. 

‘AI friends don’t behave the way that real friends do. They don’t have their own needs. They’re there to play whenever you feel like it … So how well is having an AI friend going to prepare you to go to preschool and interact with real kids?’ she said.

OpenAI has since suspended FoloToy’s access to its models, while the manufacturer has halted sales of Kumma pending a full internal safety review.

Hugo Wu, FoloToy’s marketing director, said the firm has decided to temporarily suspend sales of the affected product and begin a safety audit.

A spokesman for Miko said: ‘At Miko, children’s safety is our highest priority and we’re proud to be leading the category in child-safe use of AI. This is why we have built our own proprietary conversational AI model rather than relying on general LLM systems, which has been intentionally designed with young users in mind and enables us to carefully evaluate each response for age suitability, emotional tone and educational value.

‘Miko robots have been built by a team of parents who are experts in pediatrics, child psychology and pedagogy, all focused on supporting healthy child development and unleashing the powerful benefits responsible AI innovation can have on a child’s journey. Our products, including Miko 3, are kidSAFE+ COPPA certified and undergo annual audits and certifications by FTC-approved independent privacy programs, reaffirming our commitment to the highest standards of online privacy.

We are always expanding our internal testing, strengthening our filters, and introducing new capabilities that detect and block sensitive or unexpected topics. These new features complement our existing controls that allow parents and caregivers to identify specific topics they’d like to restrict from conversation. We will continue to invest in setting the highest standards for safe, secure and responsible AI integration for Miko products, as we strongly believe in the ability of our AI-powered products.’

The Daily Mail has approached FoloToy, Curio for comment.