Reviewed: 19 October 2026
Scope: England, with international governance context
Recent announcements about artificial intelligence in education do not point in a single direction. Taken together, however, they show a clearer pattern. Schools are being asked to manage AI as a matter of safeguarding, evidence, professional judgement, procurement and curriculum design, rather than treating it as a stand-alone technology project.
Six developments during September and October 2026 help to explain what is changing and where school leaders may need to pay attention.
1. Online safety now reaches further into technology decisions
The Department for Education has made the connection between schools’ existing online-safety responsibilities and their filtering, monitoring and cyber-security arrangements more explicit. Its standards already cover digital content in several forms, including AI-generated material, and ask schools to reconsider their safeguards when new technology is introduced.
This does not create a new AI-specific statutory duty. It does mean that adopting an AI tool cannot be separated from questions about accounts, devices, data, access controls and what staff should do when a concern arises. Our detailed review, DfE strengthens online-safety link in school technology standards, explains the distinction.
2. Government is investing in evidence, not endorsing products
England’s new four-year EdTech Testbed Programme will involve more than 1,000 schools and colleges and up to 100 rapid evaluations. The Education Endowment Foundation will invest £7 million in up to 15 more rigorous studies of promising approaches.
The important point is what the programme does not establish. It does not show that any particular technology improves learning, reduces workload or supports inclusion. It creates a route for testing those claims. For school leaders, that strengthens the case for asking what evidence applies to a particular product, group of pupils and setting before committing money or changing practice. See England launches £24 million EdTech evidence programme.
3. Young people may be confident users without being critical users
Public First’s survey of 4,021 people aged 11 to 24 in England found widespread use of AI, considerable trust in it as a way to check facts, and limited clarity about school rules. The findings concern self-reported attitudes rather than tested competence, and the study was commissioned for an advocacy report. Even with those qualifications, it identifies a practical problem: regular use can coexist with weak habits of verification.
Schools therefore need to teach more than how to operate a tool. Pupils also need to know when to question an answer, how to trace a claim to evidence, how to express uncertainty and when human expertise matters. Read New polling reveals an AI trust and guidance gap and our earlier review of Ofcom’s research on AI use and understanding.
4. AI can save teachers time, but the benefit is narrower than the headline
An independently evaluated randomised trial of Oak National Academy’s Aila lesson-planning assistant found an average saving of 49 minutes a week. The quality of the resulting resources was no worse than in the comparison group.
Adoption was limited, however. Use declined during the ten-week trial, Aila appeared in only about 7% of lessons in an average week, and the research did not measure pupil learning. The useful conclusion is restrained: a specialist AI tool can reduce planning time for some teachers, but access alone does not guarantee sustained use or better teaching. Professional checking and adaptation remain essential. Our evidence summary is AI lesson-planning trial finds time savings and uneven adoption.
5. Governance questions extend beyond privacy
UNESCO’s recent global consultation on AI in education brought together questions about age appropriateness, learner safety, procurement, total cost, digital sovereignty and explainability. It was a consultation, not a final standard or a requirement for UK schools, but its breadth is useful.
A school assessing an AI product may need to consider not only what data it collects, but also whether pupils are developmentally ready for its intended use, whether decisions can be explained, how the service may change, what dependence it creates and what happens if it is withdrawn. UNESCO’s consultation page sets out the work. Our article on turning school AI safety principles into contract terms provides a practical procurement reference point.
6. Oracy is becoming part of the response to an AI-shaped world
The Education Secretary has announced that oracy will gain equal status with reading and writing as a strand of England’s revised English curriculum. The final programme-of-study and assessment details are still to come, so schools do not yet have a new immediate requirement.
The direction is nevertheless significant. Speaking, listening, questioning and explaining give teachers more opportunities to see how a pupil reached a conclusion. Oracy is not an AI detector, but it can help make reasoning visible when a polished written answer reveals little about the understanding behind it. Read England links curriculum oracy reform to the rise of AI and Assessment debate shifts from AI detection to visible reasoning.
What school leaders can do now
- Map current use. Establish which AI tools are already being used by staff and pupils, on which devices and with what accounts or data.
- Join the relevant decisions. Bring safeguarding, curriculum, IT, data protection and procurement perspectives together before adopting a tool.
- Separate different kinds of evidence. Time saved, resource quality, teacher preference and pupil learning are distinct outcomes.
- Make expectations teachable. Give pupils clear examples of permitted use, prohibited use, disclosure, verification and responsible challenge.
- Keep human judgement visible. Use discussion, drafts, source records and short oral explanations where they help reveal the reasoning behind an answer.
The wider picture
The most consequential change is not a single tool or announcement. AI is moving into the ordinary responsibilities of running a school. Evidence, safety, purchasing, staff development and curriculum design now overlap. Schools do not need to chase every new product, but they do need a coherent way to decide what is useful, what is safe and what pupils should learn to judge for themselves.