Skip to content
All Things
Classroom
Let's talk
Blog
Policy5 min read

What UK classrooms are actually doing with AI this year.

Guidance has arrived, adoption is real, and the biggest worry has shifted from cheating to data. A field guide for school leaders and parents.

Varchas KoushikVarchas KoushikFounder · Storyteller & Learning Experience Designer

If you lead a school, or your child is in one, it is hard to get a plain picture of what AI use looks like right now — as opposed to what a vendor or a headline says it looks like. Third Space Learning's 2026 report on AI in UK schools is one of the more grounded accounts, and a few of its themes are worth pulling out.

1. The Department for Education has drawn a line

The DfE has published product safety expectations for generative AI tools used with pupils. This matters more than it sounds. It gives schools something concrete to hold suppliers to — filtering, age-appropriateness, how data is handled — instead of relying on a demo and a good feeling. If a tool can't meet those expectations, that is the conversation over.

2. Most adoption is behind the scenes

The report describes AI use as predominantly administrative: lesson planning, resource creation, exam preparation. Live classroom use is much rarer. That is not a failure — it's the sensible first step — but it does mean that most pupils' direct experience of AI is happening at home, unsupervised, rather than at school with a teacher.

3. Pupils are ahead of policy

Which brings us to the finding that should worry leaders most: a large share of children already use generative AI for schoolwork, often before their school has a usage policy. The gap between what students do and what schools have decided is where academic integrity problems, privacy problems and simple bad habits grow. A policy that arrives a year late is a policy that arrives after the habits have formed.

4. The top concern is now data, not cheating

School leaders in the report name data protection as their primary worry — particularly free tools that may train on what pupils type into them. That is the right instinct. It is also the strongest argument for teaching AI literacy rather than just managing AI access: a student who understands that a chatbot conversation can be stored and reused will make better decisions everywhere, not only at school.

What to do with all this

  • Write the policy now, in plain language, even if it's short. Revise it in a term.
  • Train staff beyond admin use — the classroom is where the interesting questions live.
  • Teach pupils explicitly. Verification, privacy and judgement don't arrive on their own.
  • Hold every tool to the DfE's safety expectations before it meets a child.
Varchas Koushik

Written by

Varchas Koushik

Founder · Storyteller & Learning Experience Designer

Founder of All Things Classroom. 10,000+ hours facilitating sessions with students across age groups and geographies; writes about game-based learning, story-driven teaching and what's possible when a lesson is designed like an experience instead of a script.

More about Varchas