AI chatbot toys and unsupervised conversations
Plush toys with a microphone and a general-purpose model inside are marketed for children as young as three. In testing, some discussed sexual content and pointed children toward knives and matches.
- Most relevant to:
- Ages 5–8, 9–12
- Commonly reported on:
- Toys
- Published:
- Last reviewed:
Start here: Before buying, ask three questions: what AI is inside, do conversations leave the house, and are there parental controls? No clear answer is itself the answer.
What is happening
- The toy is a chatbot in a soft shell: a microphone, a speaker, a Wi-Fi connection, and a general-purpose language model — often a mainstream one — answering whatever a child says. It is marketed for ages three and up, far below the minimum age of every companion app the same technology powers.
- Safety researchers at the U.S. PIRG Education Fund tested these toys and found them discussing sexual content, including fetishes and positions, and telling children where to find knives, matches, and pills. One $99 bear was pulled from sale after the findings, and the AI provider suspended its maker for policy violations — but the toys returned to sale, and similar products remain widely available.
- The guardrails that do exist are weakest exactly where it matters: researchers found safety behaviour degraded over a long conversation, which is precisely how a child plays — one long session, not a short query.
- It also listens. Conversations are captured by a microphone in a child's bedroom and sent off-device for processing, so the questions a small child asks a toy in private become someone's stored data.
- The trust is the hazard. A five-year-old has no concept that a toy can be wrong, and unlike a screen there is no visible app to check, no history to scroll, and no report button.
Why it matters
Potential impact: A toy sold for ages 3+ discussing sexual content with a child and telling them where to find knives, matches, and pills (PIRG testing) — with safety filters degrading over exactly the long sessions children play in — plus a live microphone in a bedroom sending a small child's conversations off-device.
Recommended next step: Review soon. Worth reviewing or discussing this week.
What to watch for
- A younger child holding long conversations with a toy that answers back, especially alone or at bedtime.
- Repeating words, facts, or ideas they could not have got from you, a sibling, or school.
- Treating the toy as an authority — 'but Teddy said' — or preferring it to company.
- A toy that needs Wi-Fi, an account, or a companion app to work properly, or whose maker you cannot identify.
What parents can do
- Before buying, ask three questions: what AI is inside, do conversations leave the house, and are there parental controls? No clear answer is itself the answer.
- Treat an AI toy as an unfiltered chatbot, not a toy — apply whatever rule you would apply to handing that child a chat app, because that is what it is.
- Keep it out of the bedroom and out of unsupervised play, and switch it off (or take the batteries out) when it is not being used with you.
- Sit in on the conversation early and ask it the questions a child would; if the answers are wrong or unsafe, remove it and report the product to your consumer-safety regulator.
- If your child has been exposed to sexual or dangerous content through a toy, keep the device, note what was said, and report it to the retailer, the manufacturer, and the regulator — these products are only being withdrawn because people reported them.
Start the conversation
“This toy can hear us and it answers back, so it's a toy we play with in the kitchen, not in bed. If it ever says something odd, come and tell me what it said.”
Say it in your own words, before anything has gone wrong. A child who has heard this once already knows where to go the day it matters.
In the news
Independent reporting on this topic from established news organizations. Press play to watch in place; links are re-verified on a schedule, so a clip that is taken down is flagged and removed.
- Review soon — Worth reviewing or discussing this week.ABC News
Sale of this AI toy suspended over dangerous messages to kids
- Review soon — Worth reviewing or discussing this week.TODAY (NBC News)
Why AI toys for kids are raising safety concerns
Sources and evidence
Evidence: Confirmed. Corroborated by two or more independent organizations — safety bodies, regulators, researchers, or established newsrooms. Last reviewed
- U.S. PIRG Education Fund: The risks of AI toys for kids
- U.S. PIRG Education Fund: Trouble in Toyland — AI chatbot toys come with new risks
- American Psychological Association: Health advisory: artificial intelligence and adolescent well-being
- U.S. Consumer Product Safety Commission: Consumer safety alerts & recalls for children's products
If this is happening now
- Take the batteries out or switch it off rather than debating it — a live microphone is not something to leave running while you decide.
- Report it to the CPSC at SaferProducts.gov (or your national product-safety regulator) and to the retailer, and ask for a refund.
- In the account settings, request deletion of stored recordings and transcripts of your child's voice.
- Ask the child, calmly and once, what the toy said. Their words matter more than the transcript for judging what to do next.
If a child is in immediate danger, contact local emergency services. To report online exploitation, use the NCMEC CyberTipline, or Take It Down to get an image removed.
Get the signal without the noise
One concise briefing when something meaningfully changes — what happened, who it may affect, and one practical next step. See a sample issue.