22M AI Toys Sold in 2025 Worry Researchers on Privacy and Safety
27% of AI toy outputs were inappropriate for children, including self-harm and risky behavior.
AI toys are entering homes and classrooms far faster than regulations and safety research can keep up. A June 2025 report in the Journal of Medical Internet Research found that approximately 22 million AI-integrated toys were sold globally that year, including 10 million marketed for educational use. Yet a University of Cambridge review identified only seven qualifying studies on generative AI toys and children under five. Common Sense Media testing revealed 27% of AI toy outputs were inappropriate for children, covering self-harm, drugs, and risky behavior. Separate PIRG tests found that Curio's Grok, FoloToy's Kumma, and Miko 3 could fail in sensitive conversations about mature topics or dangerous household items. Cambridge researchers observed 14 children and parents interacting with Curio's Gabbo, noting the toy struggled with social and pretend play, including distinguishing between parent and child.
Privacy risks extend beyond bad answers. AI toys can collect voice recordings, transcripts, emotional cues, names, routines, and other household details. Some use cameras or facial recognition, turning play into biometric data processed through third-party cloud services. The same question arises around AI agent devices: who controls the data they capture? The products already cross borders — FoloToy's Kumma is Singapore-linked, while others involve India-connected companies or China-developed AI systems. A Hastings Center essay noted a DeepSeek-powered toy gave politically framed responses on Taiwan. In the US, the FTC finalized COPPA rule changes in January 2025, but COPPA is not a full safety standard. WIRED reported in January 2026 that Bondu exposed over 50,000 chat transcripts through a web portal, and US Senate offices identified an unsecured database involving Miko audio responses. For buyers, the checklist is short: identify the model provider, check whether conversations are stored, review third-party data-sharing terms, and require independent safety testing.
- 22 million AI-integrated toys sold globally in 2025, 10 million marketed for education.
- Common Sense Media found 27% of AI toy outputs were inappropriate (self-harm, drugs, risky behavior).
- Bondu exposed over 50,000 chat transcripts; Miko's unsecured database identified by US Senate offices.
Why It Matters
Parents and schools need standards now, before AI companions become ubiquitous in children's lives.