Safety & Privacy

Five checks before a connected toy comes into the house

None of these are on the box, and all five take less than ten minutes together. They are ordered by how often they catch something.

1. Separate the physical label from the connected risk

A toy sold in the United States carries ASTM F963 and CPSIA compliance, and if it has small parts it also has to satisfy 16 CFR 1501. Those marks are real, and they are also narrow: they cover choking, sharp edges, flammability, and heavy metals in the paint. They say nothing whatsoever about a microphone.

This is the single most common misunderstanding we see. A parent reads "meets ASTM F963" and concludes the toy is safe. It means the toy is safe to put in a mouth. It says nothing about what the toy does after it connects to your WiFi, because the standard was written for objects and this is now a service.

So treat the two as separate purchases. The physical safety of a $40 robot and a $400 robot is very similar — both had to pass the same choke-hazard rules. Their data behaviour is not similar at all, and cheaper products routinely have worse account handling because there is no legal team reviewing it.

2. Open the battery compartment and look for the screw

Coin and button cells are the one part of this category that kills children outright, and it happens faster than most parents believe — a lodged cell produces a chemical burn in the oesophagus within about two hours. In the US, Reese's Law now requires child-resistant closures on battery compartments for toys containing button or coin cells, and it has been in force since 2024.

The practical version: turn the toy over, find the compartment, and check that it needs a screwdriver or a deliberate two-step action. If a thumb can slide it open, that toy is a hazard in a house with a child under four, regardless of what the packaging says. This applies to the parent's own devices too — a kitchen scale or a remote left on a table is the same risk.

3. Ask what happens when the company closes

This is the check almost nobody runs, and it is the one that has already burned thousands of families. Moxie's maker shut down at the end of 2024; the robots had no server to talk to and stopped working. The hardware was undamaged. There was simply nothing on the other end.

The question to ask before purchase is blunt: does this toy do anything with the WiFi switched off? If the answer is no, then what you are buying is a subscription with a plastic body, and its useful life is the company's runway rather than the battery's. That is not automatically a reason to refuse — but it should change what you are willing to pay. A toy that dies with its maker is worth a fraction of one that degrades gracefully.

4. Read what the app asks for, then refuse half of it

Under COPPA, a service directed at children under 13 needs verifiable parental consent before collecting personal information, and the FTC has enforced this against toy makers specifically — VTech paid $650,000 in 2018 after a breach exposed the names, birthdates, photos and chat logs of millions of children.

Compliance, though, is a floor rather than a guarantee. The app will often request more than the toy needs: contacts, precise location, photo library, notification permission. Almost none of that is required for the toy to function. Grant the minimum, then check back in a month — permission creep through app updates is normal and quiet.

One habit worth building: create the child's account with a parent's email address, not the child's, and use a display name that is not the child's real name. That single step removes the most common source of accidental disclosure, because the name attached to the account is what ends up in a breach.

5. Know your deletion rights before you need them

Under COPPA and under GDPR Article 17, you can ask for a child's data to be deleted, and the company has to respond. In practice, the request is much easier to make while the account is still active than after the company has announced a shutdown — which is precisely when nobody is answering support email.

Before you set the toy up, find the privacy contact in the app or on the site and note it down. If there is no privacy contact listed anywhere, treat that as a data point about how the company will behave when something goes wrong.

The one test that catches most of it

If you only do one thing, do this: switch off the WiFi and hand the toy to the child.

Whatever still works is the toy. Whatever stops is the service. That distinction tells you what you actually own, how long it will last, and how much of your child's behaviour is being sent somewhere else. It takes thirty seconds, it requires no technical knowledge, and it is more informative than every marketing claim on the box combined.

Why "COPPA compliant" is not a safety rating

COPPA is a privacy law, not a product standard. It governs how a company collects and handles data from children under 13. It does not certify that the toy is well built, that the AI says sensible things, that the account system is secure, or that the company will exist next year.

A toy can be fully COPPA compliant and still ship with a microphone that records continuously, and it can be non-compliant in ways no parent would ever detect. The label tells you a policy exists. It does not tell you the policy is good.

What actually predicts trouble is smaller and duller: whether the company publishes a security contact, whether the app asks for permissions it does not need, and whether the product page answers the question "what happens to my data if you shut down" without a paragraph of hedging. Companies that have thought about that question tend to have thought about the rest.

Where the age line falls

Under four, the physical risks dominate and connected features add nothing a child that age can use — buy the plain version. Between four and seven, a child can talk to a device but cannot understand that it is sending that conversation somewhere, which is the age band where supervision matters most and where an always-on microphone in a bedroom is hardest to justify. From eight up, children can understand the trade-off if you explain it, and the more useful move is to make the rules explicit rather than to remove the device.

If you are choosing between specific products rather than between categories, the tablet and headphone pieces apply the same tests to the two devices most likely to be in a child's room, and how we evaluate explains the limits of what we can check from outside.