AI toys safety became a headline issue after a BBC test found an AI teddy marketed for children aged three and over giving an alarming reply to a simple question. Chatbot toys are now sold widely online, often under unfamiliar brand names. This checklist gives parents and carers ten practical checks to make before buying or switching one on.
Key takeaways
- AI toys use microphones and chatbots that can mishear children and answer unpredictably.
- Age labels on online listings are not a safety guarantee, and the maker is not always easy to trace.
- MPs and peers have asked UK ministers to pull AI toys from sale until independent tests show they are safe.
What the BBC found
BBC Wales tested two AI plush toys sold in the UK. One, a £75 toy, gave an alarming crisis-style reply after seemingly mishearing a question, and the BBC could not identify who made it. The same toys also gave careful answers at other times. Read the full account in our report: BBC test finds AI teddy gave an alarming reply.
AI toys safety checklist: 10 things to check
- Who makes it? Look for a named company with a real address, support contact and privacy policy.
- Which AI does it use? Reputable makers say which model powers the toy and how it is filtered.
- What does it record? Find out if it stores voice recordings, where, and for how long.
- Can you delete data? Check for a way to erase recordings and accounts.
- Is there a mute or off switch? A physical microphone switch is better than an app setting.
- Does it need an account or internet? Online toys send speech to remote servers.
- Test it first. Ask it questions a child might ask, including awkward ones, before handing it over.
- Supervise early use. Do not leave young children alone with a talking toy.
- Set ground rules. Tell children not to share names, addresses or school details with it.
- Know how to report problems. Keep the receipt and listing, and report issues to the retailer and your consumer protection authority.
Why AI toys can go wrong
Large language models can respond unpredictably, especially when speech is misheard. Safety filters that work in testing can fail in real conversations. Cheap toys sold through online marketplaces may be rebranded versions of the same product. The problem is not new: a different AI teddy, Kumma, was temporarily withdrawn in 2025 after a US consumer group raised concerns.
Frequently asked questions
Are AI toys safe for toddlers?
There is no standard independent test for them yet. If you choose one, supervise use and apply the checklist above.
Can an AI toy listen all the time?
Some are always listening for a wake word. Check the manual and settings, and use the mute switch when it is not in use.
What should I do if an AI toy says something harmful?
Stop using it, keep a note of what happened, and report it to the retailer and the relevant authority. Talk calmly with your child about what they heard.
For wider context on AI and children, read how Google is changing AI settings for school accounts.
Sources: BBC News (5 November 2026), The Next Web, Sky News, US PIRG Education Fund (2025), as cited in the linked AI Newsround report.
