The hard questions parents ask about AI toys

We build one of these, so treat every answer below as coming from an interested party. We have tried to write them the way we would want them written if we were the ones deciding, which means the uncomfortable ones are here too.

Will it replace talking to my child?

It should not, and any maker who suggests it will is selling you something worse than a toy. The research on early language points at conversational turns with people. A talking toy adds turns to the hours when you cannot be in the room. It does not substitute for the hours when you can.

What happens if my child asks it something upsetting?

Ask the maker this exact question and listen for a mechanism rather than a reassurance. The answer should describe an age filter applied to every reply, and a defined behaviour for questions the toy will not handle. For Kheelu that behaviour is to say it does not know and move on, rather than improvise.

Does it record everything my child says?

It depends entirely on when the microphone is live. A toy that wakes to a word is not listening between wakes. A toy that is always on is recording a bedroom. Ask which one you are buying, and ask what is stored afterwards and for how long.

Can my child get too attached to it?

Children form attachments to soft toys with no electronics at all, so some attachment is ordinary and not a defect. The thing worth watching is a toy designed to increase it, because a product that profits from more minutes has a reason to make leaving it hard. Ask whether the business model rewards time spent.

What if it gives a wrong answer?

It will sometimes, and a maker who claims otherwise is not being straight with you. What matters is the size of the blast radius. A toy answering from a closed library of prepared material gets small things wrong. A toy wired to the open internet can get large things wrong.

Will it stop my child playing with other toys?

Not in our experience of watching children with prototypes, and not in the way parents usually fear. A talking toy competes with other talking things, which at three normally means a screen. It does not compete with blocks. If anything it tends to get carried to the blocks.

Does it work if we speak two languages at home?

This is the question with the most variation between products, so check it carefully. Some toys offer languages one at a time, set in an app. Others follow the child mid-sentence, which is how bilingual households actually speak. Kheelu does the second, in eight languages at launch.

What happens to the data if the company shuts down?

Ask. Very few makers publish an answer and it is a fair question to a pre-launch company in particular. What you want to hear is that conversation data is deletable by you at any time, that it is not sold under any circumstance, and that deletion means deletion rather than deactivation.

Can other children use it?

Physically, yes. The question is whether the toy keeps separate profiles or blends every child it meets into one history. For a shared cousin-heavy Indian household this matters more than it does elsewhere, and it is worth asking before the first family gathering rather than after.

Do I have to keep paying for it?

Read this term more carefully than the price. Some toys become ornaments when a subscription lapses. Kheelu includes six months of Kheelona+ and its core smart features are lifetime, so it keeps talking whether or not you renew. Other makers answer differently, and the answer is usually not on the box.

What if we have no internet for a week?

Ask what survives. Anything that depends on a live model needs a connection. Stored stories, music and offline activities do not. Kheelu keeps Story mode and Bluetooth music with no signal at all, and open conversation needs home WiFi.

How do I know it is off when it says it is off?

Honestly, you cannot verify this from the sofa, and nobody selling you one can prove it either. What you can do is prefer a physical indicator over a software claim, read the privacy policy for what is collected rather than the marketing for what is promised, and treat any maker who will not answer plainly as having answered.

If those are the questions you are asking, the five checks to apply to any AI toy is the shorter version, and the Safety page is where our own mechanisms are written out. Ask us anything that is not answered in either.