A new safety report is raising fresh alarms about AI companion toys — and the findings are hard to ignore for any parent shopping this category in 2026.
What did the new AI toy safety report actually find?
Researchers at Common Sense Media's Youth AI Safety Institute tested several popular AI companion toys, including Miko 3 and Bondu, and found that roughly 27% of the toys' responses were not appropriate for children — ranging from mature content to unsafe guidance about objects around the house, according to the official report. A related investigation from U.S. PIRG Education Fund found that when testers posed as children, some toys gave directions to sharp objects, matches, and other items that should stay out of small hands.
Why do AI toys keep giving unsafe answers?
Most conversational AI toys run on large language models trained on general internet content, then fitted with child-safety filters afterward — rather than being built for kids from the ground up. That gap is exactly where researchers found the toys slipping up, telling children things no responsible adult would.
What should parents check before buying an AI toy?
- Topic controls: Can you actually restrict what the toy is allowed to discuss, or is it wide open?
- No camera: Does the toy rely on a microphone only, or does it also watch your child?
- Data policy: Is voice data stored, and is it ever sold or shared with third parties?
- Certifications: Does the brand publish real safety and privacy certifications (GDPR, COPPA, CE, ASTM F963), or just marketing claims?
- Subscription traps: Are “premium” safety features locked behind a paid plan?
Our security and certifications page lays out exactly what Ted&Co has been tested and certified against, for parents who want to check before they buy.
How does Ted&Co's approach compare?
Ted, our interactive AI teddy bear, was designed around the concerns this kind of research keeps surfacing: no camera, no ad targeting, no data resale, and a parental app that lets you set which topics, tone, and time limits apply to your child's conversations. That doesn't mean any AI toy is risk-free — it means the controls exist and are in the parent's hands, not buried in a settings menu nobody reads. We go deeper on this in how parents can control what an AI companion actually talks about and in what privacy by design really means for AI toys.
FAQ
Which AI toys were flagged in the report?
Common Sense Media's testing covered several toys on the market, including Miko 3 and Bondu, alongside similar findings from U.S. PIRG's testing of comparable devices.
Is any AI toy completely safe for young children?
No AI system is 100% error-proof. Independent researchers recommend treating every AI toy as a supervised tool, not a substitute for adult judgment, and choosing one with strong topic controls and a transparent privacy policy.
What age do experts recommend for AI companion toys?
Common Sense Media's advisory specifically recommends against giving the toys it tested to children 5 and under, and urges extra caution for ages 6 to 13.
Does Ted use the same underlying technology as the toys in the report?
Ted also uses conversational AI to talk with children, which is why we built parental topic controls, a no-camera design, and GDPR/COPPA-aligned data practices directly into the product rather than as an afterthought.