A few years ago, 'online safety' meant social media and gaming. Now it means something else too: your child talking to an AI chatbot every evening, using AI to help with homework, or chatting with a 'companion' app that remembers their name and asks how their day went.
AI isn't a future risk to plan for. It's already part of most children's daily digital lives — and UK safeguarding organisations are only just catching up with what that means for families.
WHY AI SAFETY IS DIFFERENT FROM OTHER ONLINE RISKS
Social media and gaming risks are largely about what children see and who they meet. AI chatbots introduce something new: a relationship-like interaction that feels responsive, personal, and always available.
Research shared for Safer Internet Day 2026 by Vodafone, First News and the NSPCC found that almost a third of children who use AI chatbots see them as friends, and many confide concerns in a chatbot that they wouldn't share with a parent, teacher, or friend. NSPCC's Childline service head has said AI can offer real opportunities for children, but risks grow when children aren't sure what's real or how the technology actually works.
That combination — constant availability, a friendly tone, and no real understanding of a child's wellbeing — is why AI safety needs its own conversation in your house, separate from your existing screen time or social media rules.
THE MAIN RISKS PARENTS SHOULD UNDERSTAND
1. Emotional attachment to chatbots
Some children and teens begin treating AI companions as a substitute for real friendships, which can affect how they build actual social skills and relationships over time. A chatbot's replies are generated, not felt — it cannot recognise distress the way a trusted adult can.
2. Inappropriate or unmoderated content
Many general-purpose AI tools aren't designed with children in mind, and content filters can be inconsistent — particularly on companion apps rather than mainstream tools.
3. Privacy and data collection
Children often share far more personal detail with a chatbot than they would in a search engine — names, worries, routines, even locations — without understanding where that data goes.
4. Homework and academic dependence
AI tools can quietly do the thinking for a child rather than supporting it, which affects both learning and, in older children, academic integrity.
5. Crisis response gaps
Regulators in the UK and elsewhere have raised concerns that AI chatbots are not reliably equipped to recognise when a child is in genuine emotional distress and to direct them to appropriate human support.
6. Blurred lines around what's 'real'
Younger children especially may not distinguish between a helpful tool and something with genuine understanding or care for them.
AGE-BY-AGE GUIDANCE
Ages 5–8: Co-use only. Any AI tool should be used together, on a shared device, with you present. Avoid companion or 'friend' style apps entirely at this age.
Ages 9–12: Allow supervised, task-specific use (homework help, creative writing) with regular check-ins about what they're using and why. Keep companion-style AI apps off-limits or closely reviewed together.
Ages 13–16: Greater independence is reasonable, but keep AI use an open topic rather than a private one. Agree together which tools are allowed, and revisit the agreement as new apps appear.
Ages 16+: Focus shifts to critical thinking — helping teens evaluate AI-generated information, understand data privacy, and recognise manipulative design in companion apps.
WHAT UK PARENTS CAN DO THIS WEEK
1. Ask what AI tools your child already uses. Many children use AI chatbots without parents realising, often through apps that don't market themselves as 'AI' at all.
2. Check age ratings and settings. Most mainstream AI tools set a minimum age (commonly 13), but enforcement is weak — a support setting is not the same as a safeguard.
3. Turn on any available parental or safety controls. Where a tool offers them, enable content filters, usage limits, and activity summaries.
4. Make AI a shared topic, not a secret one. Ask what they use it for, what it's said that surprised them, and whether it's ever made them uncomfortable.
5. Set a family rule specifically for companion apps. These carry the highest risk of emotional over-reliance and deserve a clearer boundary than general AI tools.
6. Talk about what AI can't do. It can't know your child, can't feel concern for them, and shouldn't be where they turn first with something serious — that's still a trusted adult's job.
CONVERSATION STARTERS THAT ACTUALLY WORK
'What's the most useful thing that AI tool has helped you with?'
'Has it ever said something wrong or a bit strange?'
'Do you ever tell it things you wouldn't tell me? What kind of things?'
'What do you think it does with what you type?'
Open, curious questions like these get further than 'were you talking to a robot again?' — and they keep the door open for your child to come to you if something feels off.
THE BIGGER PICTURE
The UK government announced in June 2026 that major social media platforms will be required to restrict accounts for under-16s, with the law expected to take effect in spring 2027. AI safety is following a similar trajectory — regulation is catching up, but slowly. In the meantime, the most reliable safeguard in your home isn't a setting — it's an ongoing, judgement-free conversation.