Cambridge Study Warns AI Toys Misread Children’s Emotions, Urges Stronger Psychological Safety Rules

A University of Cambridge research team examined the risks of generative AI interactions with young children using Gabbo, an AI toy powered by OpenAI technology. Because children under 5 are still developing the ability to recognize and express emotions, a toy that misreads their feelings or responds inappropriately could affect their psychological safety and social learning.

The study found that Gabbo sometimes misread emotions, interrupted conversations and gave stilted responses when interacting with young children. The Cambridge team called on governments to establish “psychological safety” standards for AI products intended for children under 5. It also advised parents to allow children to use such products only in shared spaces so they can monitor the interactions throughout.