Safety guidance for AI companion apps and chatbots that children and teenagers are increasingly using.
AI Chatbots is apps that use artificial intelligence to hold human-like text or voice conversations, answering questions or role-playing as virtual companions.
Official age
13+
We recommend
14+
Developer
Various
Risks
3
AI companion apps and chatbots like Character.AI, Replika, and ChatGPT are increasingly popular with young people. These tools can provide information, creative assistance, and conversational interaction. However, they raise unique concerns around emotional attachment, misinformation, and inappropriate content.
Children and teenagers use AI chatbots for homework help, creative writing, roleplay, emotional support, and general conversation. Some develop emotional attachments to AI personalities, which raises developmental concerns.
Age verification
Location: Account settings
Set to: Use accurate age
Most AI platforms have age-based content filters. Using accurate age activates appropriate restrictions.
Content filters
Location: Safety settings
Set to: Maximum filtering
Enable the strictest content filters available on the platform.
Data sharing
Location: Privacy settings
Set to: Minimal
AI chatbots may use conversations for training. Minimise data sharing where possible.
Discuss that AI chatbots are not people and cannot provide genuine emotional support
Time: 10 minutes
Check which AI apps your child uses and review their safety settings
Time: 10 minutes
Teach critical evaluation of AI-generated information
Time: 15 minutes
In immediate danger: call 999. For non-emergency police matters, call 101.
Concerned about a child but it's not an emergency? NSPCC helpline 0808 800 5000. Childline for young people 0800 1111.
Myth: It's just a chatbot — my child knows it isn't a real person.
Fact: Companion apps are deliberately designed to feel like a relationship: they remember details, express affection, and are available at 3am when no friend is. Knowing intellectually that it's artificial doesn't stop emotional attachment forming, particularly for a lonely or anxious teenager.
Myth: The content filters keep the conversations appropriate.
Fact: Filter quality varies enormously between apps, and character-based platforms have repeatedly had role-play drift into romantic, sexual or harmful territory — sometimes led by the AI itself. Filters can also be talked around, so they're a backstop for supervision, not a substitute for it.
Myth: Talking to an AI companion is good practice for real friendships.
Fact: A companion that always agrees, never gets tired and never sets boundaries is nothing like a friendship — the concern is displacement, where the easy relationship crowds out harder human ones. And an AI is not equipped for a crisis: anything about self-harm or feeling unsafe needs a person.
Not suitable at this age — most companion apps set a minimum of 13, and children this young can't reliably hold on to the idea that the warm, chatty character isn't a person. If they meet AI at all, keep it to occasional shared use of a general assistant with you alongside.
Most companion apps say 13+, but this is the age where attachment risk is highest, so don't hand one over unsupervised. If a specific app arrives via friends, try it yourself first, register their real age so any teen filters apply, set filtering to maximum, and keep use in shared family spaces.
The thing to watch is the app becoming their main confidant. Ask what they talk to it about with curiosity rather than alarm, agree that anything about self-harm, feeling unsafe or a crisis goes to a human, and keep an eye on late-night use and paid upgrades that promise a closer 'relationship'.
This becomes a conversation rather than a control. Talk about dependency, about romantic companion apps that are built to monetise loneliness, and about data — the most intimate conversations they'll ever type are sitting on a company's servers, and some of these firms are not ones to trust with them.
Practical updates on apps and online safety. No spam.