2026-08-28 · Sienna Richardson

AI Chatbots and Kids: A Parent's Guide

Two things changed, and most advice has not caught up

Your child isn't just watching content anymore. They're talking to it - and the ground under this topic moved twice in the last nine months.

In November 2025, Character.AI removed open-ended chat for all users under 18. That was the platform's core feature and the subject of nearly every parenting warning written about it. In August 2026, OpenAI switched on ChatGPT for Teens globally, applying additional content protections automatically to accounts identified as belonging to under-18s.

Both changes are real improvements, and both mean that articles written before them - including most of what currently ranks for these searches - describe products that no longer work the way they say.

What did not change is the shape of the risk. It simply moved. The large platforms with legal exposure have invested heavily in child safety; the demand for unrestricted AI companionship did not disappear, it relocated to smaller apps with no safety team, no age enforcement and, in several cases, no content filtering at all. The app you have heard of is now rarely the one worth worrying about.

What Kids Are Doing with AI Chatbots

Homework and schoolwork

This is the obvious use. Kids ask ChatGPT to explain math concepts, help with essays, or generate study guides. On its own, this is fine. The concern is that some kids have stopped learning entirely and just copy-paste AI responses as their own work.

Creating characters and stories

Character.AI lets users create and talk to fictional characters, and children build them from the shows, books and games they like. Creative, and for a while the single biggest source of concern in this category.

As of 25 November 2025 this is closed to under-18 accounts: Character.AI removed open-ended chat for users it identifies as minors. The caveat is that the age is self-declared, so a child who entered an adult birthdate at signup still has an adult account. Checking the birthdate is now worth more than any setting. The full picture is here.

Emotional support

This is where it gets complicated, and it is also the part with the best evidence behind it.

Common Sense Media surveyed 1,060 US teenagers aged 13 to 17 in April and May 2025, with the fieldwork run by NORC at the University of Chicago. The findings are worth reading in full, but four numbers carry most of it:

72% of teens have used an AI companion at least once
52% use one regularly, meaning at least a few times a month
A third have chosen an AI companion over a person for a serious conversation
Nearly a third found AI conversations as satisfying as, or more satisfying than, human ones

The third and fourth numbers are the ones that should hold your attention. The first two describe a popular product; those two describe substitution. Common Sense Media's own recommendation off the back of this research is that nobody under 18 should use AI companions at all.

For lonely, anxious or socially isolated children, an AI companion fills a real void. It is also not equipped to handle a child in crisis, and the ease of the relationship is precisely what makes it displace the harder human ones.

Inappropriate content

Despite safety filters, kids have found ways to generate explicit, violent, or otherwise inappropriate content from AI chatbots. This is especially true on platforms with fewer guardrails than ChatGPT. Some smaller chatbot apps have essentially no content filtering at all.

The Risks Parents Should Know About

Emotional dependency

A chatbot that's always available, never judges, and always says what you want to hear is designed to be addictive. Kids (and adults) can develop unhealthy attachments. The chatbot becomes a crutch that prevents developing real coping skills and real relationships.

Privacy

When your child tells a chatbot about their day, their fears, their relationships, and their personal information, that data goes somewhere. Most chatbot companies store conversations. Some use them for training. Your child may be sharing deeply personal information with a company they've never heard of.

Inappropriate content generation

AI chatbot safety filters are imperfect. Kids actively test them. Jailbreaking techniques spread rapidly through social media and school group chats. A filter that works today may not work after your child watches a TikTok about a new bypass.

Academic dishonesty

This is a school problem as much as a home problem, but parents should be aware. If your child's grades suddenly improved and they can't explain their own homework when asked, AI might be involved.

What Parents Can Do

1. Know which platforms exist

The landscape changes quickly, and the useful distinction is not which app is most popular but which has anyone accountable for child safety.

Platform	Under-18 policy	Parent controls	Filtering	Where the risk sits
ChatGPT	Teen protections automatic since Aug 2026	Yes, no transcripts	Strong	Wrong answers, schoolwork
Google Gemini	13+ via Family Link	Yes, via Family Link	Strong	Wrong answers, schoolwork
Claude	Built for 18+	No	Strong	Not intended for children
Character.AI	Open-ended chat banned for under-18s	No	Strong since the ban	Fake ages on old accounts
Replika	18+	No	Limited	Inappropriate for a minor
Chai	Nominally 18+, weakly enforced	No	Minimal	Explicit content, dependency
Janitor AI	Nominally 18+, weakly enforced	No	Effectively none	Explicit content, dependency

The bottom three rows are where the genuine problems now are, and they are precisely the names a parent will not recognise on a screen time report. New ones appear continuously, which is the central practical difficulty in this whole category.

2. Talk about it openly

Banning AI entirely is impractical and will be increasingly so as these tools become integrated into school and work. Instead, have an honest conversation about what's appropriate and what isn't.

"Are you using any AI chatbots? Which ones? What do you use them for? What kinds of things do you talk to them about?"

You might be surprised how willing your child is to talk about this. For many kids, AI chatbots are just another app. They don't think of the conversations as private the way they think of texts with friends.

3. Set clear rules

AI is fine for learning but not for doing homework you turn in as your own
Don't share personal information (full name, address, school, phone number) with any chatbot
If a conversation makes you uncomfortable, close it and tell me about it
Some platforms are off limits (parents should review each platform's safety record)

4. Control access at the computer level

AI chatbot sites can be blocked like any other, and the domain list and the four methods are here.

The structural problem is that a blocklist has to have heard of an app before it can stop it, and this is the fastest-moving category on the internet. Allowlisting inverts that: with 3Eyes, every site is blocked until you approve it, so you can allow chatgpt.com for homework while every companion app launching next month is already blocked without you knowing its name.

The other half is what the machine offers instead. A child using AI to avoid learning needs something that teaches rather than something that stops - 3Eyes carries educational games, learning activities and a built-in tutor that works through a topic rather than handing over an answer. Setup is about two minutes.

5. Monitor usage patterns

You don't need to read your child's chatbot conversations. But you should know how much time they're spending with AI chatbots. If your child is spending 3 hours a day talking to Character.AI, that's a sign that something in their social life needs attention.

Looking Ahead

AI chatbots aren't going away. They're going to become more capable, more lifelike, and more integrated into daily life. The kids growing up right now will use AI tools throughout their education and careers.

The goal isn't to prevent your child from ever interacting with AI. It's to make sure they develop a healthy relationship with it: using it as a tool without depending on it emotionally, sharing information wisely, and understanding the difference between an AI that's programmed to be agreeable and a real person who cares about them.

This is new territory for every family. Nobody has all the answers. But being aware of the issue and having conversations about it puts you ahead of most parents.

Frequently asked questions

What age can a child use an AI chatbot? Most major platforms set 13 as the minimum and expect parental permission under 18. Below 13 the practical problem is not explicit content but that a young child cannot tell a confident wrong answer from a right one.

Can I read my child's AI conversations? On the major platforms, no. ChatGPT's parental controls provide restrictions and safety notifications but not transcripts, and Character.AI has no parent-facing controls at all. If reading conversations is your goal, no mainstream AI platform offers it.

Which AI apps should worry me most? Not the famous ones. Apps with no age enforcement, no safety team and no meaningful filtering carry nearly all the residual risk, and they are the ones you will not recognise.

Is AI making my child worse at schoolwork? It can, and the mechanism is specific: asking for an explanation is learning, asking for the answer is not, and the two look identical from outside. One explicit rule - ask it to explain, don't ask it to write - covers most of this.

Should I ban AI entirely? For under-13s, reasonable. For teenagers, rarely effective: it relocates usage to a phone or a friend's laptop, and these tools are becoming part of how school and work operate. Rules about use, plus blocking the unreviewed companion apps specifically, gets further.