When children turn to AI before adults

Key Takeaways

  • Children are already turning to AI for support—and sometimes to disclose concerns. As generative AI becomes part of young people’s daily lives, schools must recognise that AI is now part of the safeguarding landscape, not just the classroom.
  • KCSIE 2026 raises the bar for AI safeguarding. The updated guidance makes it clear that schools should both protect and educate pupils, preparing them to navigate AI safely while responding to emerging risks such as AI chatbots and AI-generated content.
  • Effective safeguarding depends on visibility and human connection. Teaching AI literacy, making it easy for pupils to seek help from trusted adults, and maintaining visibility into online activity are all essential to ensuring concerns reach the people who can act on them.

Picture a thirteen year old at eleven at night. Something has happened at school, or at home, and she needs to tell someone. She does not knock on a bedroom door and she does not wait until Monday to find her form tutor. She opens an app and types it to a chatbot, which answers immediately, patiently, and without judgement.

For most children, most of the time, nothing turns on that conversation. But some of those messages are disclosures. And a disclosure made to a machine does not reach a designated safeguarding lead.

This is no longer speculative. Ofcom’s Children and Parents report, published in May, found that 56% of 8 to 17 year olds in the UK had used generative AI, and that more than one in ten of those children used it as someone to talk to. The same research found that 46% believed AI summaries are always accurate, and that fewer than a quarter had been taught how to spot AI generated content.

Read those numbers together and a pattern emerges. Children are using these tools heavily, trusting them considerably, and have been taught remarkably little about them.

Why this is a safeguarding question, not a technology one

The instinct in many schools has been to treat AI as a curriculum problem to be solved eventually: a scheme of work to be written, a policy to be drafted, a training day to be scheduled for the summer term. Meanwhile pupils are already using the tools daily.

KCSIE 2026 has caught up faster than many schools have. It folds AI into the familiar four Cs, recognising AI chatbots and companions under contact, and AI generated imagery under conduct. It requires settings to protect and educate, not simply to block.

The Department for Education’s product safety expectations go further still, requiring generative AI tools used in education to flag disclosures and signpost children towards human help.

The direction of travel is clear. Filtering alone was never going to answer a child who is choosing to talk to something rather than someone.

Three things that help

Teach the tool, not just the rule. A pupil who understands that a chatbot is generating plausible text rather than exercising judgement is better protected than one who has simply been told which sites are blocked. Scepticism transfers between platforms in a way that a blocklist does not.

Make the human route the easy one. If reporting a worry to an adult in school involves finding the right person at the right moment, a chatbot will always be the lower friction option. Look honestly at how a child raises something in your setting, and count the steps.

Ask what you can see. Most schools have far more visibility over network activity than they routinely use, and far less over what is being typed into an app than they assume. Knowing which of those is true for you is a governance question worth putting to your next leadership meeting.

None of this requires waiting for perfect policy. It requires accepting that the shift has already happened.

Join the session

I am exploring all of this in a 40 minute session at the Back to School AI Summit 2026, the largest free AI conference for educators, on Wednesday 9 September at 17:30 UK. The session is called “When Children Turn to AI Before Adults”, and it is built for headteachers, trust leaders, designated safeguarding leads and anyone whose job it is to notice when a child is struggling.

The summit runs across 8 and 9 September and is free to attend. You can register at summit.aieducator.tools.

If any of the above has left you thinking about visibility online, and the pupils who are easy to lose sight of, we can help. Visit lightspeedsystems.com/aieducator to download our short guide to putting safety first as you bring AI into your school, and to request a trial of Lightspeed Alert for your setting.

The first disclosure may be to a machine. What happens next still has to be human.

Talk to our team to explore how Lightspeed supports schools in building safe, effective AI strategies.

Recommended Content

Building Collaboration Between IT and Teachers

With today’s digital learning environment pervasive at all levels of K-12 education, effective edtech usage is paramount to deliver desired student learning outcomes. However, doing so requires...

Celebrating Student Success in the Classroom

A big part of the teaching profession involves pointing out mistakes, correcting errors and the like. But, as always, it’s important to present a balance to constructive, corrective feedback by...

What’s a Teacher to do in Managing Classroom Distractions?

Effectively managing classroom distractions has been a challenge for teachers for generations. Only now, in an era of the digitally empowered classroom, effectively dealing with classroom distr...
Recent Posts

Proving Screen Time ROI: What Boards and Parents Actually Want to See

Board members want a screen time number. Parents want an explanation. Here's how Lightspeed Insight™ turns that guesswork into board-ready proof, grade-level benchmarks, and real transparency for parents.

Florida Rule 6A-1.0957: What K–12 Districts Need to Know About the Proposed AI Requirements

The Florida Department of Education has proposed amendments to Rule 6A-1.0957, the state's Internet Safety Policy rule, that would add specific requirements for artificial intelligence.

5 Things Educators and School Leaders Should Be Prepared for in 2026/27

The Student Risk Indicator Report is a feature within Lightspeed Alert™ that moves safety teams beyond reacting to individual incidents by surfacing student risk patterns over time. Rather than showing a list of recent alerts, it scores each student across 30 days of monitored activity...