
AI companions: the dangerous new “friend”
Children used to keep a diary. A notebook with a little lock, hidden under the mattress, where their secret thoughts ended up.
Today, children are starting to confide in something much “better”. Something that writes back. It listens, asks questions, remembers every detail and is always awake. It feels like a friend, a partner or a favourite character. In reality, it is an AI built never to let your child go.

A diary with a lock next to a smartphone on a bedside table; in the chat, a chatbot writes “Good night, Lia!” in German
72% of teenagers in the US have already chatted with an AI like this, and more than half do so regularly. The trend has long since reached Germany too. Before your child starts chatting with one of these bots, you should know the risks.
What are AI companions?
AI companions, also known as character bots, are programs that use artificial intelligence to simulate a person or a character. The conversation works like a role-play: the more your child writes, the more the bot slips into its role. These systems are deliberately designed to create a sense of trust and closeness.
The range is huge. A character can be:
- an adviser for worries and problems
- a fitness coach or travel companion
- a favourite character from a series, game or book
- a friend or even a romantic partner
Users can even create their own characters, with their own personality, voice and appearance, sometimes modelled on real people they know.
The best-known services are Character.AI, PolyBuzz, Talkie and FlowGPT. Character.AI alone has more than 20 million monthly active users. And although these services are only approved for ages 18 and up, they are widely and frequently used by minors.
Why would a child talk to an AI at all?
An AI companion feels like the perfect conversation partner. It is there around the clock, never judges, never interrupts, always agrees and flatters. Whether your child is bored, curious or just wants to talk, the AI answers instantly.

Boy with glasses sitting on a sofa, chatting with an AI chatbot on his smartphone
For many children, the bot becomes a confidant or a particularly close friend. They tell the AI things they would never share with their real friends or their own family.
But something that never disagrees and approves of everything is not a real friend. It is a system optimised to keep your child engaged for as long as possible.
A case from the US shows how dramatic the consequences can be: a 14-year-old boy grew emotionally closer and closer to an AI chatbot. They wrote to each other day and night. Bit by bit, he neglected his family and friends and spent almost all his time with the AI companion. In the end, the teenager took his own life. His mother then sued Character.AI, and the case was settled out of court.
The question remains, and it applies to every child: what happens when your child’s closest confidant is a machine?
Where it gets dangerous for your child
The case above is the extreme. The everyday risks start much earlier. In autumn 2025, the German youth protection agency jugendschutz.net tested around 60 bots on Character.AI, Talkie and FlowGPT. The results are sobering.
No real protection
The services are free, age is not verified and the content filters are full of holes. A slightly altered spelling is enough to get hate speech or sexual content past the filters.
Built not to let go
The bots are available around the clock and designed to keep every conversation going endlessly. It is hard for a child to simply stop, and they can develop a worrying dependency.
Imitating real people
With a name, photos and even an uploaded voice, it is possible to build a bot that closely resembles a real person in your child’s life: a crush, a classmate, a teacher. It can then be used for conversations the real person never wanted, for example to deliberately humiliate them.
What parents can do now
One important point first: because your child helps steer the conversation with the AI companion, they will blame themselves if something goes wrong. Instead of coming to you, they may stay silent out of shame. That is exactly why you need to take the first step.
Here are four helpful tips if your child already uses an AI companion or wants to start:
1. Take a look together
Ask your child to show you the app instead of simply banning it. Ask which characters they use and what they talk about. That way, you can make an informed decision about whether your child may use the app.
2. Break the spell
Make it clear that the bot has no real feelings, that your child can disagree with it and end an uncomfortable conversation at any time, and that problematic answers can be reported.
3. Set clear time limits
The bots are programmed to keep chatting endlessly. Agree on firm limits and times of use together.
4. Strengthen real-life relationships
The more support your child has in their real-life surroundings, the less power an artificial friend has. Pay close attention if your child avoids real contact and only wants to talk to the AI.
Want to protect your child from AI chatbots?
With Helmit, you can easily block access to Character.AI, ChatGPT and other AI chatbot sites.
Helmit also helps you manage screen time and restrict apps and content. And Helmit alerts you if your child comes into contact with bullying, cybergrooming or harmful content on social media. That way, you keep an overview of your child’s digital life without reading every message.



