AI chatbots and kids: a UK parent's guide
By Lewis C. Based on each company's own guidance and Internet Matters research, checked 4 October 2026. Written from the official guides, not yet checked on my own devices. Spotted something?
Quick answer
Most UK kids already use them. In Internet Matters' survey, 64% of children had used an AI chatbot, and over a third of those said it was like talking to a friend. The main ones have a minimum age of 13. ChatGPT and Snapchat's My AI have parent settings worth using; Meta AI in WhatsApp can't be switched off at all. The settings matter less than one message, repeated: it's a computer, it gets things wrong, and real worries go to a real person.
The apps, the age rules and the parent settings
| App | Minimum age | What parents can do |
|---|---|---|
| ChatGPT | 13, with a parent's permission under 18 | Link your account to theirs (both have to agree). You can set quiet hours and turn off memory, voice and image making. Since August 2026, under-18s get teen protections automatically. You can't read their chats. |
| Snapchat My AI | 13 (Snapchat's own minimum) | In Family Centre, go to Settings and turn on Disable My AI. My AI then won't reply to them or process what they send. |
| Meta AI (WhatsApp, Instagram, Messenger) | 13 | You can't switch Meta AI off. On Instagram, parent supervision can block one-to-one chats with AI characters, rolling out in the UK in 2026: check the Supervision settings in your app. |
| Character.AI | 18 for chatting | Since 25 November 2025, under-18s can't have open-ended chats with its characters. They get a "Stories" mode instead. |
| Google Gemini | Not on UK child accounts | Google says Gemini isn't available to supervised (Family Link) accounts in the UK. Teens with their own Google account are a different matter. |
What can go wrong
- Treating it like a friend. 35% of children who use chatbots told Internet Matters it's like talking to a friend, and six in ten parents worry their kids think chatbots are real people. Vulnerable children were nearly three times as likely to use companion bots like Character.AI.
- Advice on big things. 23% had asked a chatbot for advice, from what to wear up to mental health. Chatbots can sound sure of themselves and still be wrong.
- Romantic and sexual roleplay. It's the reason Character.AI shut chat off for under-18s. ChatGPT's teen protections and Meta's teen settings now restrict it, but teens who lie about their age skip those.
- Made-up answers in homework. Chatbots invent facts and sources, and they say them confidently.
- Personal details and photos. Anything typed or uploaded can be stored. ChatGPT now warns teens before they upload a sensitive image.
What to do
- Ask which ones they use. It's often more than you think, because Snapchat and WhatsApp have one built in.
- Use the settings that exist: link ChatGPT accounts, turn on Disable My AI in Snapchat's Family Centre, and check Instagram's supervision settings.
- Set one house rule for homework. Mine would be: chatbots can explain and quiz, but the work they hand in is theirs.
- Say the line out loud: "It's a computer. It can be wrong, it doesn't care about you, and it's not a person you can trust with secrets."
- Point real worries somewhere real. Childline is free and confidential on 0800 1111, for kids who don't want to talk to you yet.
What the law says
In February 2026 the government announced it would bring AI chatbots, including ChatGPT, Gemini and Copilot, under the Online Safety Act's illegal content duties. Until then, chatbots that only let you talk to the bot itself sat outside much of the Act. The minimum ages above are the companies' own rules; there's no UK law setting an age for chatbots. Social media is different: see the UK age rules and the planned under-16 ban.
Questions parents ask
What age can kids use ChatGPT?
13, and under-18s need a parent's permission. Since August 2026, ChatGPT applies teen protections automatically when it thinks a user is under 18, and parents can link accounts to set quiet hours and switch features off.
Can I turn off Meta AI in WhatsApp?
No. Meta doesn't let you remove it. Teach your child it's a computer, not a person, and not to tell it anything they wouldn't tell a stranger.
Is Character.AI safe for kids?
Since 25 November 2025, Character.AI hasn't allowed under-18s to chat with its characters. If your child is using it to chat, they've said they're 18.
Are AI chatbots covered by UK online safety law?
Partly. In February 2026 the government announced it would bring chatbots like ChatGPT, Gemini and Copilot under the Online Safety Act's illegal content duties, closing a gap that left some of them out.
Where this comes from
- Internet Matters: Me, Myself and AI (children and chatbots research)
- OpenAI Help: managing parental controls in ChatGPT and ChatGPT for Teens
- TechCrunch: OpenAI launches ChatGPT for Teens (18 Aug 2026)
- Snapchat Support: how to limit your teen's access to My AI
- Character.AI: under-18 chat announcement and TechCrunch on Stories (25 Nov 2025)
- Google For Families Help: Gemini and supervised accounts (not available in the UK)
- Mobile Ecosystem Forum: Meta's parental AI controls on Instagram (US, UK, Canada, Australia in 2026)
- CNBC: AI chatbot firms brought under UK online safety law (16 Feb 2026)