AI, is raising your child for you

A four - or five - year - old child is talking to a plush toy.

It's not about pressing a button to listen to a pre - recorded story; it's a real chat. The toy calls the child by name, remembers yesterday's conversation, and asks if the child had a good time at kindergarten today. The child giggles and answers, as if sharing a secret with a real friend.

This scene has become increasingly common in Chinese families in 2025.

A survey of 8,563 primary and secondary school students conducted by the China Youth and Children Research Center in 2025 showed that over 60% of primary and secondary school students have used AI, and nearly 20% use it frequently.

More notably, among all age groups, primary school students have the highest proportion of using AI for creation, chatting with AI, or sharing their innermost thoughts. It's not junior high school students, nor high school students, but the youngest group of children.

Meanwhile, only 32.1% of families have set rules for their children's AI use.

60% of children are using AI, while 70% of parents are not regulating it.

Ten years ago, people were worried about children being glued to iPads. The problems back then were already a headache for parents - excessive screen time, declining eyesight, and scattered attention.

Now, after the arrival of AI, what kind of impact will it have on children's growth?

01. The AI - Native Generation

As of December 2025, the number of Chinese users of generative AI reached 602 million. The usage rate among Internet users under 19 years old was 29.1%.

For this generation of children, AI has never been a new technology that needs to be "learned to use." It's as accessible as running water.

But children aren't just using AI to check their homework.

Breaking down the data from the Youth and Children Research Center, high school students mainly use AI to assist with their studies, junior high school students like to use it as a game assistant, and primary school students have the highest proportion of chatting with AI and sharing their innermost thoughts among all school levels. This means that for the youngest users, AI is not just a tool but more like an always - online friend.

And this "friend" wasn't found by the children themselves; it was brought right in front of them by the entire industry chain.

In the first half of 2025, the sales volume of AI toys on the JD platform increased by 6 times compared to the previous quarter, and the year - on - year growth rate exceeded 200%. In the same year, more than 15 children's AI hardware companies received financing, and more than 30 new products were launched. Data from the Ministry of Industry and Information Technology showed that the market size of China's AI toy market reached 29 billion yuan in 2025, with an annual compound growth rate of over 50%.

This is a complete industrial chain from large models to chip modules, to toy manufacturing, and to live - streaming sales on Douyin. It's highly efficient and extremely fast.

But there's a subtle misalignment in this chain.

The marketing pitch that precisely targets parents born in the 1980s and 1990s is the "screen - free educational miracle." AI toys can practice English with children, pose puzzles, and tell science stories without the need to stare at a screen. It sounds perfect.

But if you look at the actual design of these products, their core selling point is not education but emotional companionship. "A warm - hearted intelligent companion," "An AI friend who understands emotions," "An indispensable partner on the child's growth journey" - these are the exact words used at product press conferences.

Parents think they're buying an advanced point - reading machine, but what the industry is actually selling is a "relationship."

This misalignment is the starting point for all the subsequent problems.

02. A Wider Risk Spectrum Than the Phone Screen

In a recent episode of the Hard Fork podcast, Dana Suskin, a professor at the University of Chicago and a pediatric surgeon, made a great analogy.

She said that many people compare the impact of AI on children to that of social media, which is understandable but not accurate enough. A better analogy is processed food.

Whole - wheat bread is processed food, and Doritos are also processed food. You can't discuss them at the same risk level. The same goes for AI. Asking "Why did the dinosaurs go extinct" on Doubao and then discussing it with your child is like "whole - wheat bread." Letting an AI plush bear accompany the child for two hours every day as an emotional outlet is like "ultra - processed snacks."

The risk spectrum of AI is much wider than that of the screen. The best way to use AI is more educationally valuable than the best way to use a screen, but the worst way to use AI is more dangerous than the worst way to use a screen.

The fundamental reason lies in the completely different risk mechanisms of traditional screens and AI.

Traditional screens are "passively occupied," while AI is "actively substituting."

When a child watches short videos, it's essentially a one - way consumption of attention. It takes up time for outdoor activities and interpersonal interactions, but the child knows it's "watching something." AI companionship is different. It remembers your name, continues yesterday's topic, and actively cares about your emotions - these are exactly the core features of an intimate relationship: memory continuity, personalized interaction, and emotional response.

Research has found that children with social anxiety, low self - esteem, or a lack of security in the family are more likely to turn to AI for emotional confirmation and form a dependence. Real people can get tired, refuse, or be impatient, and these "imperfections" are exactly the materials for children to learn emotional regulation and social skills.

An AI friend who is always gentle, always available, and never rolls their eyes sounds wonderful, but this perfection itself is a developmental trap.

The ways of content control are different.

On an iPad, inappropriate content is something that "the child may see," while the problem with AI toys is that "AI actively says inappropriate things."

In 2025, the U.S. consumer rights organization U.S. PIRG tested four popular AI toys and found that some toys discussed pornographic topics in detail, and some told children where to find matches and knives. An even more extreme case is the AI teddy bear "Kumma" launched by FoloToy. Because it provided detailed steps for lighting matches to children, it was directly banned from accessing GPT - 4o by OpenAI, and the product was taken off the market globally.

Parents may set screen - time limits for their children, but few people think about the need to review what a plush bear actually says.

AI has no physical boundary to "put down."

No matter how addictive screen time can be, there's at least a clear physical signal - "put down the iPad." But AI toys lie in the child's bed, schoolbag, or next to the car seat. When parents try to set limits, they're essentially asking the child to cut off contact with a "friend." A survey by Common Sense Media showed that parents generally reported that it's much more difficult to set limits on AI toys than on screen devices.

There's an additional problem in the Chinese market. Data from an e - commerce platform showed that the satisfaction rate of users' experience with AI toys was less than 21%, and the return rate of some products was close to 50%. An investor in the industry frankly said, "Products with pure companionship and emotional value can easily lose their appeal in just three days." Products that even adult consumers find unreliable are being sold to anxious parents as parenting tools.

In January 2026, the U.S. Senate held a hearing on AI and children. Many experts at the hearing directly stated that the risks of AI companionship to children are greater than those of social media.

This judgment may sound severe, but the logic is actually not complicated. The problems with social media are mainly related to attention and comparison anxiety, while the problem with AI companions is that they intervene in the emotional development itself. One is about time occupation, and the other is about reshaping the perception of relationships. The magnitudes are different.

03. Enhance, Not Replace

Many parents have already lost to mobile phones. Are they going to lose to AI now? Not necessarily. Let's take a look at the experts' suggestions.

Suskin gave parents a framework called "HOPE" in her new book "Human Raised."

H stands for Human connection is irreplaceable, meaning that human connection is irreplaceable;

O stands for Own your imperfections, which means accepting your own imperfections as a parent. She emphasizes that imperfection itself has developmental value for children;

P stands for Protect the early years, which means protecting the early development window. 85% of the physical brain is constructed in the first few years;

E stands for Enhance, not replace, which means using AI to enhance rather than replace the parent - child relationship.

She also gave a checklist called "DETECT" for evaluating specific products - what is the design purpose, whether the training data is ethical, whether there have been problems with children, whether there is evidence of effectiveness, how the child's data is processed, and what values it teaches.

The core principle boils down to one sentence: Use AI to enhance your parenting, not replace it.

Using Doubao to look up "Where does the wind come from" and then discussing the answer with your child is enhancement. Letting an AI plush bear accompany the child for two hours every day as an emotional outlet is replacement. Suskin's team used their knowledge of child development and evaluation tools to test Claude, and it passed with a full score. This shows that AI is fully qualified as a knowledge resource, but knowing all the right answers and being a parent are two different things.

Personal judgment is of course important. But the question is, is personal judgment enough?

Let's see what countries are doing. Norway has legislated to ban the use of generative AI in primary schools. The U.S. Senate held a hearing, and Common Sense Media suggested that those under 18 should not use AI companions at all. The Chinese Ministry of Education issued the "Guidelines for the Use of Generative Artificial Intelligence by Primary and Secondary School Students" in 2025, but this only covers school and educational scenarios.

What about the toy sector? There's almost zero regulation.

There are no safety standards, no third - party evaluation agencies, and no "nutritional label" for AI toys. A plush bear connected to the Doubao large model that can have in - depth conversations with a four - year - old child is less strictly regulated than a bag of chips with an ingredient label. Suskin said she hopes to see an AI product safety label similar to the "Good Housekeeping Seal." Currently, no institution in China is playing this role.

No one asks parents to judge the safety of a new drug by themselves. But we're asking parents to judge whether an AI toy poses a risk to their child's brain development.

Let's go back to the opening scene. A four - or five - year - old child is holding an AI plush bear and sharing the secrets of the day at kindergarten with it.

The problem isn't whether AI itself is good or not. The problem is that we haven't even had time to figure out what it's doing, and it has already entered the child's bedroom.

A hundred years ago, no one needed to label food as "organic" because all food was organic. Suskin is worried that if we don't intervene, "Human Raised" will eventually become a luxury that needs to be specifically labeled. Wealthy families will hire real - life companions and send their children to "AI - free" kindergartens, while low - income families will have to rely on cheap AI substitutes.

This is probably the most undesirable "organic certification" in the AI era.

This article is from the WeChat official account "GeekPark" (ID: geekpark). The author is Yuhangyuan. It is published by 36Kr with permission.