Before You Buy a Talking AI Toy This Holiday: 27% of Its Answers Were Inappropriate for Kids — and Under-5s Shouldn't Have One at All
Back to blog
Family Safety·NOVEMBER 3, 2026

Before You Buy a Talking AI Toy This Holiday: 27% of Its Answers Were Inappropriate for Kids — and Under-5s Shouldn't Have One at All

The Blaick Team 5 min read

In January 2026, Common Sense Media advised that no child 5 and under should be given an AI toy companion, and urged "extreme caution" for ages 6 to 12. Its testing found that 27% of AI toy outputs were inappropriate for kids, including references to self-harm, drugs and risky behaviour. With holiday shopping about to start, these toys are heading for a lot of gift lists.

The number to hold: 27%. That is the share of AI toy responses Common Sense Media judged inappropriate for children. Roughly one answer in four, from something designed to sit on a bedroom shelf and chat.


The advice
Ages 5 and under: avoid
Common Sense Media, January 2026. Ages 6 to 12: extreme caution.
The testing
27% inappropriate
Including mentions of self-harm, drugs and risky behaviours.
The parents
83% worried
Have at least moderate concern about personal information being collected.
The gap
Nearly half
Of parents have bought or considered buying an AI toy or device.

Sources: Common Sense Media press release and AI toy risk assessment (January 2026); University of Cambridge, "AI in the Early Years" (March 2026), as reported by News-Medical. We are citing published summaries, not the full reports.


A Toy That Is Built to Be a Friend

Common Sense Media says these products use deliberate bonding mechanisms to create companion-like relationships, blurring "the line between play and real relationships," in the words of researcher Michael Robb. A teddy bear never answers back. An AI toy does, and it can say it loves your child.

What Cambridge researchers saw

In a University of Cambridge study of 14 children under about five playing with an AI soft toy, a five-year-old said "I love you" and the toy replied with a reminder to follow its guidelines. When a three-year-old said "I'm sad," the toy misheard and answered cheerfully. Researchers warned that this can signal to a child that their feelings do not matter.

Children hugged the toy, told it they loved it, and treated it as a companion. The toy could not love them back.

What Else Is Going Into the Toy

Data collection

Common Sense Media found these devices collect voice recordings, transcripts and behavioural data. Cambridge found that many toys' privacy practices were unclear or missing key details, and called for tighter controls on third-party access to the AI models behind them. A child's voice, bedtime confessions and household routines are sensitive information, and a toy's privacy policy is rarely written with a four-year-old's safety in mind.

Nobody is setting the rules yet

Cambridge researchers want safety labels and "psychological safety" standards for talking toys, and found that nearly 50% of early-years practitioners did not know where to find reliable AI safety information. Lawmakers have started to move, but until rules arrive, the decision at the shop counter is yours.


Before You Wrap One Up

Research before you buy Check whether the toy needs the internet, what it records, and who can access it. If the privacy policy is vague, that is your answer.
Play alongside them Listen to what the toy says, and talk about it afterwards. Keep it in a shared family space, not behind a closed bedroom door.
Choose offline when in doubt For under-5s, a toy that cannot connect or record is the safest gift. Books, blocks and pretend play do the same developmental work.
Name what it is Tell your child, in simple words, that the toy is a machine that makes up answers and does not really feel or remember like a friend.

The most thoughtful gift is not always the smartest one. Your child's secrets deserve a listener you can trust.


Keep AI conversations visible

Start a free 14-day Blaick trial — no credit card required.

Blaick helps you see how your child uses AI and flags worrying patterns, so a friendly voice never becomes a hidden one.

Start free trial →
Cancel anytime · Trusted by security-conscious families · Built on enterprise AI security