The Education for a Connected World framework has given schools a clear, age-related structure for teaching children about online life. But the digital world has changed substantially since the current edition was published. Generative AI, synthetic media, automated decision-making and algorithmic recommendation now influence how children learn, socialise and make sense of the world.
Why now?
AI is already being used to plan lessons, create resources, support revision and generate text, images, audio and video. Used well, it can reduce routine workload and open new creative possibilities. Used uncritically, it can encourage cognitive offloading, expose personal data, reproduce bias, generate convincing errors and weaken the link between effort, understanding and achievement.
At the same time, the regulatory landscape is shifting. Ofcom’s child-safety duties under the Online Safety Act 2023 now require regulated services to assess and reduce risks to children. Ofcom specifically identifies recommender systems, adult-to-child contact features and designs that encourage extended use as potential risk factors. Its current register also recognises depression content and body-stigma content as forms of non-designated content that may cause significant harm.
The Government has also announced plans to prevent under-16s from using certain social-media services, with the first restrictions expected in spring 2027. Regulations are still being developed and will require parliamentary approval. This is an important distinction: professionals should prepare for the direction of travel without presenting every implementation detail as settled.
Infographic 01
The changing landscape
A curriculum response must keep pace with law, platform design and classroom use.
The aim is not to remove every risk from children’s digital lives. It is to reduce harm by strengthening understanding, judgement and agency.
What should underpin an AI-literacy strand?
The starting point must remain human. AI is an enabler—not a substitute for the teacher, the trusted adult, professional judgement or the developmental value of children thinking for themselves.
Infographic 02
Five foundations for AI literacy
Keep professional judgement in the loop
Tools such as Oak National Academy’s Aila show how AI can help teachers build and adapt resources while keeping the professional in control. That final point matters. AI output may sound authoritative while still being inaccurate, biased, unsuitable or fabricated. The person using it—and the organisation approving it—retains responsibility.
Protect personal data
Schools should use approved tools, understand how data is processed and avoid placing identifiable pupil information into public generative-AI systems. The DfE’s current guidance gives a useful practical example: generate the structure of a communication without personal data, then add the pupil’s name and class locally. Privacy notices should also explain any wider collection of information such as IP address, location, browser or system data.
Teach emotional as well as technical literacy
AI literacy cannot stop at prompts and productivity. Children also need to recognise when a system is designed to hold attention, simulate intimacy or become an emotional substitute. This is not about demonising technology. It is about helping children retain perspective, seek human support and understand the commercial and behavioural forces shaping their experience.
What could children learn?
A new strand should be progressive. It should move from recognising how systems shape experience to applying ethical and professional judgement independently.
Infographic 03
A developmental journey
Illustrative outcomes—not a complete curriculum.
I can…
- explain why recommender systems prioritise some content;
- spot signs that an image, voice or video may be AI-generated;
- prompt safely without sharing personal information;
- recognise body-stigma, depressive and other harmful content;
- use reporting, blocking and support routes;
- explain the difference between risk and harm;
- describe how age assurance is intended to protect children.
I can…
- evaluate outputs for bias, error and hallucination;
- check sources and identify synthetic or manipulated media;
- explain authorship, copyright and intellectual-property concerns;
- use AI for routine support while retaining responsibility;
- follow assessment rules and acknowledge permitted AI use;
- evaluate a platform’s safety controls and age checks;
- make reasoned choices about when not to use AI.
Academic integrity must be explicit. JCQ requires students to submit their own independent work. Unacknowledged AI-generated or AI-paraphrased material may constitute malpractice. Schools therefore need clear, regularly reviewed expectations—not vague warnings about “cheating”.
How should settings implement it?
This is not the responsibility of one computing teacher or one annual safety lesson. A meaningful AI-literacy strand needs a whole-setting approach connecting curriculum, safeguarding, data protection, technical standards, staff development and leadership oversight.
Map existing curriculum, policies and staff practice against the knowledge children now need.
Define which tools, users, data and purposes are acceptable—and where AI should not be used.
Use the four DfE modules, with Module 3 as the essential safety foundation for all staff.
Use discussion and diagnostic tools such as ProjectEVOLVE Knowledge Maps to identify gaps and misconceptions.
Make expectations clear and listen to children’s real experiences rather than relying on adult assumptions.
Measure changes in judgement, culture and behaviour—not simply how many lessons or resources were delivered.
Investment in connectivity can support this work, but access alone is not digital inclusion. Children also need suitable devices, skilled adults, accessible design and opportunities to connect away from screens through sport, creativity, nature and community. Digital agency grows best within a balanced childhood.
Useful resources
- Education for a Connected WorldUKCIS age-related digital knowledge and skills framework
- ProjectEVOLVE ToolkitFree resources mapped to the framework
- ProjectEVOLVE Knowledge MapsExplore understanding, gaps and misconceptions
- Ofcom child-protection dutiesCurrent regulatory guidance under the Online Safety Act
- DfE: Generative AI in educationPolicy position and expectations for settings
- DfE AI product safety standardsCapabilities and safeguards expected in education products
- DfE AI training modulesFour free modules plus leadership materials
- AI and data protection in schoolsPractical DfE guidance for protecting personal data
- JCQ: AI use in assessmentsRequirements for independent work and preventing malpractice
- Filtering and monitoring standardWhole-setting responsibilities for effective safeguarding
- Planned social-media restrictionsGovernment fact sheet on the expected 2027 changes
- Oak National Academy: AilaAn example of teacher-controlled AI lesson support
From “don’t share” to “know, judge and act”
The AI revolution does not require us to discard good online-safety education. It requires us to extend it. A ninth strand would give schools a coherent way to build children’s knowledge over time: how AI works, how it influences them, how to use it safely and creatively, when to question it—and when to step away.
That is the shift we now need: from reactive restriction to informed agency; from access to capability; and from technology-led change to human-centred education.
Discuss this workstreamAccuracy note: Regulatory information was reviewed on 16 July 2026. The proposed under-16 social-media restrictions remain subject to detailed regulations and parliamentary approval. Settings should check the linked official sources for updates.