AI is becoming common in UK schools, showing up in lessons, marking, planning, and some government pilot programs. Some classrooms use chatbots to tutor pupils; others rely on adaptive platforms to set practise and spot gaps. But questions about data, exam integrity and who pays are never far away. Here’s a look at where things stand in 2026, what’s changed recently, how the industry is reacting, and what schools might expect going forward.
Quick reference
Snapshot for busy leaders:
- Government Pioneer Group: up to eight suppliers, £300,000 each for design/testing; target reach 450,000 pupils by end of 2026, according to government announcements.
- Common tools in classroom: ChatGPT (ChatGPT Plus ~£16/month for individual accounts in 2026), Google Workspace for Education Fundamentals (free), Microsoft 365 A1 for Education (free tier available).
- Typical procurement costs: free core suites; light AI licences £1–£5 per pupil/yr; dedicated tutoring services or one-to-one sessions cost tens of pounds per session depending on model.
- Top risks: pupil data protection (UK GDPR), exam malpractice, bias and hallucination in AI outputs, unequal access between schools.
Current state — how UK schools are using AI in 2026
AI is now embedded across routine school activity. Teachers use generative tools to draft lesson plans and learning objectives. Adaptive platforms set and mark practice tasks, adjusting difficulty as pupils progress. Chatbots are used for low-stakes tutoring and language support in multilingual classrooms. Senior leaders use AI to model timetables, forecast staffing needs and steer intervention budgets.
How schools use AI depends on their level and type. Primary schools favour text-to-speech, reading-level re-leveling and simple adaptive readers. Secondary schools adopt more sophisticated tutoring pilots for GCSE and vocational courses. Special schools use AI-driven personalised programmes for communication and engagement.
Key developments since 2024
Two big changes have made a real difference. First, central government pushed pilots for AI tutoring: the DfE invited EdTech and AI firms to join a Pioneer Group, offering up to £300,000 each for co-design and testing, with an aim of reaching as many as 450,000 pupils by the end of 2026. The funding supported a few UK and international companies to run classroom trials for Year 9 and 10 students.
Sure, second, regulators sharpened guidance. The Information Commissioner's Office updated advice on children’s data and artificial intelligence, stressing data minimisation, model risk assessment and the use of data protection impact assessments (DPIAs) before deployment. Exam regulators likewise set clearer policy: schools must not permit unauthorised use of generative AI in assessed work.
Top picks and analysis — what schools are actually buying
There are four common buying categories:
- Core productivity suites — Google Workspace for Education and Microsoft 365 A1 remain widespread. Both have free tiers suitable for many schools; their AI add‑ons vary by licence.
- Generative chatbots — ChatGPT is the most visible tool in corridors and staff rooms. ChatGPT Plus attracts teachers and parents at around £16/month for personal accounts; schools buying enterprise contracts often negotiate different rates and data controls.
- Adaptive learning platforms — suppliers such as Century Tech and Sparx Maths continue to supply curriculum-aligned adaptive practice, with licences often bought at a school or trust level. Those platforms now include explainable-AI dashboards for teachers.
- AI tutoring pilots — the DfE Pioneer Group pilots include purpose-built tutoring systems designed to map to the national curriculum for Year 9–10 pupils. Those pilots aim to prove impact for disadvantaged pupils before wider rollout.
Comparison table: common tools, cost signals and classroom fit
| Tool category | Example products | Price guide (2026) | Best for |
|---|---|---|---|
| Productivity suites | Google Workspace for Education Fundamentals; Microsoft 365 A1 | Fundamentals/A1: free (core features) | Whole-school email, docs, basic collaboration |
| Generative chatbots | ChatGPT (OpenAI), Google Gemini | ChatGPT Plus approx. £16/month (individual); enterprise pricing variable | Lesson drafting, pupil prompts, language support |
| Adaptive practice | Century Tech, Sparx Maths, MyMaths | School/trust licences: typically £1–£20 per pupil/year depending on scope | Curriculum-aligned homework and formative assessment |
| AI tutoring services | DfE Pioneer Group pilots; Third-party vendors | Pilots funded (£300k design/testing grants); future costs depend on procurement | Targeted one-to-one or small-group tuition for key cohorts |
Industry impacts — workload, inequality and the market
AI has cut some teacher workload — those hours saved are often spent on small-group teaching and feedback. But the gains are uneven. Trusts with bigger budgets buy enterprise licences, legal review and training.
Smaller primaries rely on free tools and volunteer governors to run procurement.
That split threatens to widen the digital divide the government says it wants to close. The DfE’s pilot target of 450,000 pupils by the end of 2026 is aimed at narrowing that gap, but scaling proven, safe models into every classroom will be expensive and slow.
Expert views
Ian Murray, the minister for digital government, framed the pilots as a levelling tool, saying AI could deliver personalised tuition to pupils who can’t afford private help. Academics and school leaders are more cautious. Many welcome the potential to diagnose gaps faster. Yet they warn against over-reliance on opaque vendor models, and urge independent evaluation of learning impact.
Ofsted inspectors and assessment experts say the pedagogic value depends on how schools redeploy the time saved. An AI-generated lesson plan is only useful if a teacher adapts it to their class. Still, headteachers who’ve piloted tutoring systems report measurable improvements in low-stakes practice scores over a school term — when interventions are well targeted and monitored.
Practical tips for schools thinking about AI
Buy with clear requirements. Specify data residency, deletion rights and teacher control of outputs. Run a DPIA before any rollout and keep a short public privacy notice for parents.
Train staff. Offer short, practical workshops on prompt design and verification. Start with teacher-facing tools that save time, then pilot pupil-facing features in one subject for a term.
Set classroom rules. Clarify when pupils may use chatbots and what counts as homework. Use plagiarism detection and insist on transparent referencing for AI-assisted work.
Procurement and cost control: negotiate multi-year licences with clear exit clauses. Use the free tiers of core suites for universal access, and reserve paid licences for targeted tutoring or subject-specific platforms.
Privacy and safety — what governors and DSLs must check
Data protection isn't optional. The Information Commissioner’s Office expects schools to minimise pupil data shared with third-party models and to have contracts that specify lawful bases for processing. If a model is trained on pupil data, that must be explicit and subject to strong safeguards.
Safeguard mental health and safeguarding lines. AI chatbots must not replace pastoral staff. Schools should route any safeguarding disclosures made to a bot back to a human and log incidents. For younger pupils, disable free-text generation where possible and use supervised, curriculum‑mapped resources.
What's next — policy and market signals for 2027
Expect tighter procurement standards and more national frameworks. The DfE’s pilots will report outcomes through 2026; successful pilots are likely to be bundled into framework agreements that MATs and local authorities can buy from. Expect mainstream platforms to offer education‑grade, data‑protected versions of their models at different price points.
Assessment policy will firm up too. Exam regulators will publish clearer rules on permitted AI use in non‑exam assessments and coursework, and exam centres will update malpractice policies. Schools that plan now — documenting use, training staff and running DPIAs — will be ahead of regulation rather than behind it.
Related Articles
- Protect Your Data Online UK 2026 — Privacy Guide
- AI Writing Tools 2026 for Students, Bloggers, Business
- Best Smart Home Devices UK 2026 — Alexa, Google, Apple
AI in UK schools is no longer hypothetical. It's here in practical, mixed ways — saving time, offering personalised practice and raising tough questions about data and fairness. Government funding has pushed tutoring pilots into classrooms and regulators have sharpened expectations. The next 12–18 months will tell whether pilots scale into routine support for disadvantaged pupils or whether access and oversight leave too many schools behind.
This article was created with AI assistance.
