March 23, 2026, 6:32 pm | Read time: 8 minutes
Stuffed animals are daily companions for children. Whether at daycare or elementary school–at least one teddy bear is always there. But what if the stuffed animal could listen and ask questions? What once sounded like science fiction is now a reality. So-called “AI toys” are designed to enable playful learning and entertainment. But can they really promote children’s development, or do they negatively interfere with a particularly sensitive phase? A new study from Cambridge provides initial insights.
What and Why Was Investigated?
The 2026 study “AI in the Early Years” from the University of Cambridge explores a central question: What does it mean for young children when toys can talk, react, and seem like a counterpart?
The focus is on so-called GenAI toys. GenAI stands for “generative artificial intelligence.” This is a technology that responds to inputs and generates answers that resemble human language. When built into toys, it means they can talk to and interact with children. This doesn’t refer to a single type of toy but a whole range. Previous studies have examined, for example, talking educational toys or devices for reading aloud.1,2
For the practical part of the study, the researchers specifically used a toy called “Gabbo”: a soft stuffed animal with a built-in voice function. It can respond to the child, ask questions, and suggest small games–similar to a voice assistant, but in the form of a plush toy.3
Why the Early Years Are Crucial
In the early years, children develop fundamental skills: They learn to speak, understand emotions, and interact with others. Play is the most important learning space. Therefore, the researchers wanted to find out how AI toys fit into this process.
How Researchers Studied Children’s Behavior
The study combines three approaches: a systematic review of existing studies, assessments from early childhood professionals, and direct observations of children playing with an AI toy.
In the first step, the researchers conducted a systematic literature review. They searched four major scientific databases for studies on AI toys for children under five years old. The result shows how new the topic is: Worldwide, only seven suitable studies were identified. These mainly dealt with education, toy design, and user experiences–less with actual developmental impacts.
In the second step, early education professionals were involved. A total of 39 people participated in an online survey, supplemented by focus groups and a workshop. The aim was to capture experiences, expectations, and concerns from practice.
The third part provides particularly vivid insights: Here, the researchers observed 14 children aged three to five playing with a talking AI stuffed animal. This toy could ask questions, respond, and suggest small games. It was a talking stuffed animal named “Gabbo.” It is a soft plush toy with a built-in voice function that can communicate with children, ask questions, and suggest games. Subsequently, both the children and their parents were interviewed about their impressions.
It is important to note: Other, independent experts have not yet reviewed the results. Therefore, these are initial indications, but not yet definitive findings.
Also interesting: A popular activity among children is apparently dangerous for the heart
The Study’s Findings
How Children Play with AI Toys
The observations show that children interacted with the toy in very different ways. Some spoke directly to it, asked questions, or engaged in suggested games. Some children treated the toy like a character with its own preferences–for example, asking about favorite colors or considering what the toy might like to eat.
At the same time, the play situations showed that communication often did not work smoothly. In one observation, a child repeatedly answered a question while the toy was still speaking–it simply wasn’t ‘heard.’ The child then became louder and eventually tried desperately to get the plush toy to respond by shaking or shouting at it.
Misunderstandings were also common. Sometimes the toy did not respond appropriately to what the children said, or it suddenly asked a new question without addressing what was said. This led to breaks in play.
Weaknesses in Imaginative Play
A particularly important point was imaginative play–when children pretended that a toy could feel, eat, or experience something. Here, clear limits were evident. In one situation, a child wanted to pretend that the toy could sleep or open a gift. However, the toy did not engage and instead explained that it could not do such things.
This free “pretend” play is crucial for development. Children learn to understand emotions, empathize with others, and try out their own ideas. If the toy does not engage, this shared play can be interrupted.
Between Enthusiasm and Frustration
Many children visibly enjoyed the play. They found it exciting that the toy could talk and react, describing it as funny or interesting.
At the same time, there was repeated frustration. Especially when the toy did not understand them or did not respond as expected. In some cases, children ended the play prematurely or turned to other activities.
What Parents and Professionals Said
Parents particularly highlighted that the toys are interactive and can encourage children to speak. At the same time, they also expressed concerns–such as about the long-term effects of use or how children’s private data is handled.
Professionals reported significant uncertainty. Many said they did not know how to classify such toys. Frequently mentioned were a lack of information, unclear rules, and concerns about data protection and safety.
Why AI Toys Are Particularly Critical for the Development of Young Children
The results show that AI toys are not just another toy category. They intervene in a phase where children develop fundamental skills–especially in the social and emotional areas.
Opportunities–but with Limitations
Professionals see potential, particularly in the areas of language and communication. For example, the toys can stimulate conversations or provide additional linguistic incentives. However, there are also significant concerns. These include questions about data protection, content safety, and the role of such toys in children’s daily lives.
Social Development: What Machines Can’t Replace
A central point is social development. Young children learn primarily through interaction with real people–through eye contact, gestures, shared experiences, and spontaneous reactions. These aspects can only be partially replicated by AI toys.
When Toys Become “Friends”
Additionally, children may begin to form a relationship with the toy. The study described that children sometimes treated the toy like a friend or attributed characteristics to it. At the same time, there is concern that such relationships are one-sided and that children may not fully understand the difference between human and machine.
Risks in Everyday Life and Open Questions
Practical risks also play a role. Professionals expressed concern that such toys could be used as a substitute for real interaction in stressful situations–for example, when adults have little time. As a result, important social experiences could be lost.
Another critical point is dealing with sensitive topics. There is uncertainty about how a toy would react if a child shared something distressing. At the same time, it is unclear what data is collected and how it is used.
The study makes it clear that early childhood is strongly shaped by real interaction. Children learn through conversations, eye contact, and shared play. Changes in this phase can therefore have particular impacts–and this is precisely why the use of such technologies is considered particularly sensitive.
Video Game Can Detect Autism in Children with 80 Percent Accuracy
Your Child Spends Hours in Front of the TV? Here’s What You Need to Know
Assessment of the Study and Possible Limitations
The study provides important initial insights into a new research field but also has clear limitations. The overall data situation is very limited. The literature review includes only seven studies, and the observations are based on a small group of 14 children.
Additionally, the children were only observed in a single play situation. How their behavior develops over a longer period remains open.
Furthermore, parts of the research (as previously mentioned) have not yet been reviewed by other experts. At the same time, the assessments of the professionals are partly based on expectations and not solely on practical experience.
Conclusion
AI toys can talk to and interact with children–and that is what makes them so special. They open up new possibilities for play and learning. However, the study shows that this technology raises many open questions. Above all, the long-term effects on development are hardly researched so far. The central finding is therefore: AI toys have already arrived in everyday life. However, science is still at the very beginning of evaluating them.