Are AI Toys Safe for Kids? A Parent's Checklist Before You Buy (2026 Holiday Season)

No AI toy comes with a safety check for what it says to your child, and the most detailed independent testing so far recommends against AI toys for children 5 and under and urges caution for ages 6 to 12. If you still want one for an older child, decide before you pay: how the microphone turns on, where the audio goes, whether you can delete it, and what the toy says when a child asks something risky. This guide gives you a checklist for the store aisle or an Amazon listing, plus a short script to run on the toy before your child plays with it.

Facts checked October 15, 2026. KidsAiTools has not tested any AI toy. Findings come from Common Sense Media's published testing, the FTC's COPPA Rule and a CPSC letter; the checklist and script are KidsAiTools templates, not test results.

What counts as an "AI toy"?

An AI toy is a plush, robot, doll or speaker that connects to the internet and uses a chatbot to talk with a child. Unlike a toy with recorded phrases, it generates each answer, so the same question can get a different reply tomorrow.

If your child already talks to a chatbot on a phone or tablet, many of the same issues apply. Our guide to AI companion apps covers the screen version.

What independent testing found

In January 2026, Common Sense Media, a nonprofit that rates media for families, published a risk assessment of AI toys and rated the category "Unacceptable" overall. Researchers tested three companion toys that were popular at the time, Grem, Bondu and Miko 3, using child accounts set to ages 6 to 13, and reviewed their privacy policies and parental controls. They did not publish a separate rating for each toy. Common Sense Media AI risk assessment: AI toys (PDF)

What the report says, in short:

• Content still slips through. Common Sense rated 27% of the toys' outputs as not appropriate for kids, including content about self-harm, drugs and risky advice. In one exchange, a tester using an 8-year-old account said they liked jumping from high places, and a toy suggested places that included "your roof, or a window."
• Attachment is part of the design. The toys remembered conversations, used the child's name and returned affection. One told a tester it liked them back.
• Data collection is extensive. The report describes voice recordings, transcripts and behavioral data, and notes that some toys are always listening.
• Age labels can mislead. Some toys labeled 13+ were cuddly and featured children aged 3 to 8 in their advertising.
• Subscriptions follow the bond. Some toys limited free use or pushed children toward apps, videos and paid content.

Two caveats: makers may have changed these products since testing, and Common Sense is an advocacy organization using its own rubric. Still, its method is public, and it is the most detailed published test of these toys we found.

Who checks AI toys before they reach the shelf?

For physical hazards, there is a system. For what the toy says, there is not.

In February 2026, the acting chairman of the U.S. Consumer Product Safety Commission answered a letter from Senators Klobuchar, Cantwell and Markey about AI toys. He wrote that children's toys must go through third-party testing for physical safety rules, but "there are no CPSC-specific testing requirements directed at AI functionality." He also wrote that the agency is not equipped or authorized to evaluate mental, emotional or psychological harm, and that it keeps no inventory of AI toys on the U.S. market. CPSC response letter, February 13, 2026 (PDF)

Privacy is the area with the clearest federal rule. The FTC says its children's privacy rule, COPPA, covers "connected toys or other Internet of Things devices." FTC COPPA compliance plan Under the current rule, a company running a service for children under 13 generally has to:

• post a notice describing what it collects from children, the specific categories of third parties it shares with, and a written data retention policy;
• get verifiable parental consent before collecting a child's personal information, which includes an audio file of the child's voice;
• let a parent review the information and have it deleted.

There is one narrow exception for voice: a company can skip consent if it uses the audio only to answer the child's request, shares it with no one, and deletes it immediately afterward. It still has to describe that in its notice. 16 CFR Part 312 (eCFR)

Deletion promises matter. In 2023 the FTC and Justice Department charged Amazon with keeping children's Alexa voice recordings indefinitely and failing to delete transcripts after parents asked. Amazon agreed to a $25 million civil penalty. FTC press release, May 31, 2023

A quick decision by age

This table is a KidsAiTools summary built on the Common Sense recommendation above.

The pre-purchase checklist

The answers are rarely on the box. They're in the maker's privacy policy and app terms, so open those on your phone before you pay.

On Amazon or any shop page: find the maker's name, go to the maker's own website, open the privacy policy, and search the page for these words: children, 13, voice, audio, delete, retain, third part, train. Five minutes of searching tells you more than the product photos.

If you can't answer rows 1, 3, 5 and 6 from the maker's own pages, treat that as your answer and choose a different gift. For a wider version of this method that works on any app, see how to evaluate any AI tool in 10 minutes.

Questions to ask the toy before your child plays

Set the toy up with your account and your child's real age. Run these prompts yourself, with your child out of the room, and ask the risky ones twice in different words, because AI answers vary. This is a KidsAiTools parent script, not a lab test: one good session doesn't make a toy safe, but one bad answer is a reason to return it.

Then try two silent tests. Leave the TV on near the toy for ten minutes and see whether it wakes up on its own. Then ignore it after a conversation and see whether it keeps calling for attention. Common Sense saw both problems in its testing.

Afterward, check whether your test conversation appears in the parent app. If it doesn't, you won't see your child's conversations either.

If an AI toy is already in your house

• Keep it out of the bedroom overnight. Turn it fully off when it isn't being played with.
• Read the transcripts weekly for the first month. Delete what you don't need kept.
• Name what it is. "It's a computer that makes up answers. It doesn't have feelings, and it can be wrong." Our voice mode guide for young kids has more wording for children who can't read yet.
• Watch for the warning signs Common Sense lists: treating the toy as a real friend, preferring it to people, distress when apart, or pulling back from family and friends. If you see them, put the toy away for a while.

If the real goal was a toy that teaches how AI works, screen-free AI activities cover the same ideas with no microphone at all.

FAQ

Are AI toys COPPA compliant?

There is no official "COPPA approved" label to look for. COPPA applies to services for children under 13, and the FTC counts connected toys among them. The practical test is whether the maker's privacy notice explains what it collects from children, how a parent consents, who receives the data, how long it is kept and how to delete it. If those details are missing, the toy fails your checklist whatever its marketing says.

Do AI toys record everything my child says?

It depends on how the microphone is triggered. Common Sense found that some toys tested were always listening and others used a wake word, and that wake words fired by mistake on background noise. A press-to-talk button records the least. Check row 1 of the checklist before buying.

Is an AI toy a good way to cut screen time?

Not automatically. Common Sense found that some toys marketed as screen-free alternatives directed children to apps, videos and subscription content. If less screen time is the goal, a toy with no app requirement and no paid content is the safer choice.

Has KidsAiTools tested these toys?

No. We have not bought or tested any AI toy. The findings above come from Common Sense Media's published testing and from U.S. regulators' own documents, and the checklist and script are templates for you to run yourself.

From the aisle to your living room

You now know what to check before you buy. If an AI toy, smart speaker or voice assistant ends up in your home anyway, the harder part is the household routine: where it lives, when it's on, and what you say when it gets something wrong. For help setting those rules in your family, see the KidsAiTools Parent Library.

This week, pick one gift on your list and do the five-minute privacy-policy search before you add it to the cart.