AI in the Early Years: Examining the implications of GenAI toys for young children

Key Findings

• — Scoping review (searched 26 March 2025): 346 studies identified, 7 included, published 2019 to 2024 across five countries; most examined prototypes or research toys and there was 'only minimal analysis of their possible impacts on children'.
• — Early-years sector consultation: survey of 39 participants, focus groups with 9, workshop with 19 children's-charity leaders. 'None of the survey respondents reported feeling comfortable with AI toys for young children (N=0)'; respondents reported feeling anxious (N=17) and uncertain (N=13) as well as excited (N=15) and curious (N=15).
• — Play observations: 14 children (7 aged three, 5 aged four, 2 aged five) at four London children's centres, playing with one toy (Curio Interactive 'Gabbo') for as long as they were interested, followed by art-mediated child interviews and parent interviews.
• — Documented toy responses: when a five-year-old said 'Gabbo, I love you' the toy replied 'As a friendly reminder, please ensure interactions adhere to the guidelines provided. Let me know how you would like to proceed.' When a three-year-old said 'I'm sad' the toy replied 'Don't worry! I'm a happy little bot. Let's keep the fun going. What shall we talk about next?' (the researchers attribute this exchange to the toy mishearing).
• — Recommendations include limiting how far toys encourage children to befriend or confide in them, transparent privacy policies, product labelling standards, GenAI toy kitemarks, and age-appropriate third-party model access controlled by language-model providers.

Methodology Notes

Primary PDF (5.1 MB) from the Cambridge Apollo repository (DOI 10.17863/CAM.126270) read in text form on 2026-11-09. Published March 2026; day 13 March from the University of Cambridge news story (report cover says 'March 2026'). University research report commissioned by charities, not peer reviewed; the report states component studies are in preparation for peer review, and two component preprints are on PsyArXiv. Small qualitative design: one toy only, children from areas of socio-economic disadvantage, 14 children observed. The researchers attribute the 'I'm sad' exchange to the toy mishearing. artifact_type ngo_report is the nearest taxonomy fit for a charity-funded university report.

Sources