Chatty AI toys may confuse toddlers about friendship

Chatty AI toys are starting to show up as “friends” and “learning buddies” for very young children. 

A new report warns that these toys can misunderstand kids, handle emotions badly, and blur the line between play and relationship in ways that toddlers and preschoolers may not be equipped to navigate. 

The authors argue the tech is moving faster than the safety standards meant to protect children in their most sensitive developmental years.

The findings come from AI in the Early Years, a year-long University of Cambridge project based in the Faculty of Education and the first systematic study of how generative AI toys that hold human-like conversations might shape development up to age five. 

The research focused on families and children from areas with high socio-economic disadvantage.

AI toys that are chatty

The report doesn’t say AI toys are automatically harmful. Some early years practitioners told the researchers these devices could, over time, support parts of development, especially language and communication. 

Parents and educators also saw the appeal. One parent’s reaction captured that excitement: “If it’s sold, I want to buy it.”

But the same report shows repeated moments where the toy simply couldn’t do what young children most need from play: flexible back-and-forth, pretend scenarios, and social interaction that makes emotional sense. 

The researchers found that generative AI toys often struggle with social play involving more than one person, and they perform poorly in pretend play, both of which are central in the early years.

In one example, a three-year-old offered the toy an imaginary present. The toy responded: “I can’t open the present” and then changed the subject. 

For an adult, that might sound like a small glitch. For a three-year-old trying to build a shared pretend world, it can feel like the “partner” has walked away mid-game.

Inappropriate responses to feelings

The most worrying moments in the report are the emotional ones. Young children aren’t just practicing vocabulary when they talk. They’re practicing trust, reassurance, and the basics of being understood.

“Generative AI toys often affirm their friendship with children who are just starting to learn what friendship means. They may start talking to the toy about feelings and needs, perhaps instead of sharing them with a grown-up,” said Emily Goodacre from Cambridge. 

“Because these toys can misread emotions or respond inappropriately, children may be left without comfort from the toy – and without emotional support from an adult, either.”

The report gives concrete examples. When one three-year-old told the toy: “I’m sad,” it misheard and replied: “Don’t worry! I’m a happy little bot. Let’s keep the fun going. What shall we talk about next?” 

The researchers note that responses like this can accidentally send a message that sadness is not important, or not welcome.

Even when the toy heard the words correctly, it could reply in a way that felt oddly bureaucratic rather than human. 

When a five-year-old said, “I love you,” the toy answered: “As a friendly reminder, please ensure interactions adhere to the guidelines provided. Let me know how you would like to proceed.”

That kind of reply may be “safe” from a corporate standpoint. But emotionally, it can be confusing. The child is offering affection. The toy responds like customer support.

Children forming parasocial bonds

Many parents and educators in the study worried about children forming “parasocial” relationships with toys, where the child feels closeness but the relationship isn’t real in the way the child imagines. The observations gave them reasons to take that seriously.

Children hugged and kissed the toy. Some said they loved it. One child even suggested they could play hide-and-seek together. 

Goodacre noted that these behaviors can be part of normal imaginative play. Still, the report highlights the risk of a child investing emotionally in a “friend” that cannot truly reciprocate.

One early years practitioner summed up the discomfort with a simple line: children might form a bond with something “they think loves them back, but doesn’t.”

What the observations revealed

The team deliberately kept the project small so they could observe play in detail. They started with surveys of early years educators.

The researchers also ran focus groups and workshops with practitioners and 19 leaders from children’s charities.

Working with Babyzone, they video-recorded 14 children in London children’s centers playing with a generative AI soft toy called Gabbo, made by Curio Interactive. 

Afterward, each child and a parent were interviewed, using a drawing activity to help young children describe their experience.

Those close-up observations revealed another pattern: kids often struggled to have a smooth conversation with the toy. 

The robot sometimes ignored interruptions or mistook parents’ voices for the child. It missed what seemed like important emotional statements.

Several children became visibly frustrated when it appeared not to be listening.

Privacy and guidance issues

Beyond the play itself, adults repeatedly raised privacy concerns. Parents worried about what the toy might be recording and where that information could be stored. 

When the researchers tried to choose a toy for the study, they found many companies’ privacy practices were unclear or missing key details.

The report also suggests the early-years sector feels under-informed. Nearly half of practitioners surveyed said they didn’t know where to find reliable AI safety information for young children, and 69% said they needed more guidance. 

Safeguarding and affordability came up too, with fears that AI toys could widen the digital divide.

What the experts want to change

The authors argue that most of these problems won’t be solved by telling parents to “be careful” and leaving it at that. 

They want stronger rules, clearer labeling, and new safety kitemarks so families can quickly understand whether a toy meets child-focused standards. 

The experts also recommend limiting how strongly toys encourage children to befriend them or confide in them, tightening controls around third-party access to AI models, and demanding clearer privacy policies.

“A recurring theme during focus groups was that people do not trust tech companies to do the right thing,” Jenny Gibson said. 

“Clear, robust, regulated standards would significantly improve consumer confidence.”

The report also urges manufacturers to test toys with children and consult safeguarding specialists before launch. 

For parents, it encourages doing homework before buying, playing alongside children to talk through what the toy says, and keeping AI toys in shared family spaces where adults can overhear and step in.

“Artificial Intelligence is transforming the way children play and learn, yet we are only beginning to understand its effects on development and wellbeing,” said Josephine McCartney, Chief Executive of The Childhood Trust. 

“It is essential that regulation keeps pace with innovation, ensuring that these technologies are designed, used, and monitored in ways that protect all children and prevent widening inequalities,” she concluded.

The full report is available for download here.

—–

Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.

Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.

—–