Read Ofcom’s children-and-AI findings with their age ranges, survey populations and methodological caveats attached. The May 2026 release is not new October research, and population statistics cannot establish an individual child’s habits. Use the original report to check what each number actually measures.
A Research Release Is Not Today’s Behaviour
Ofcom published its children’s media findings on 21 May 2026. This explainer revisits that evidence, checked on 8 October; it does not announce a new survey. Before sharing a headline about young people and AI, separate the publication date from when the information was collected. A report can help you frame a question without describing what every child is doing this week.
Keep The Denominator Attached
The report says 56% of 8–17-year-olds had used AI. Its 11% companionship finding concerns AI users aged 8–17, not every child. Those populations are different. Copy the age range and the group beside any percentage you record. Otherwise a cautious finding can become a much broader claim as it travels from a report to a headline and then into a family conversation.
Check The Method Before Comparing Years
Ofcom changed its survey methodology and questionnaire, introducing face-to-face interviews alongside an online panel and a modular design. It describes trends over time as indicative. Treat a year-to-year comparison cautiously rather than attributing every difference to changed behaviour. For the details, read the report’s methodology on printed page 9 and its AI section on printed pages 26–31. Keep the caveat beside the finding when sharing it.
Make A Five-Line Evidence Card
Clued suggests a short exercise before using any statistic in a presentation, school discussion or family decision. Write five lines: the original source; the publication date; who was asked; the exact question or measure; and what the result cannot tell you. Leave an unanswered line blank rather than guessing. A missing detail is a reason to investigate, not something to conceal behind a confident summary. Then ask a second reader to explain your card back to you. If they describe a different population or a stronger conclusion, revise the wording. You are practising careful communication, not testing that reader’s intelligence. Save the original link so another person can inspect the same evidence rather than relying on your paraphrase. For a classroom exercise, compare two summaries of the same source. Highlight the words that widen the claim: all, always, proves or causes. Check whether the original evidence supports each one. Then choose the summary that preserves the question, population and uncertainty most faithfully. This is an editorial reading exercise; it is not a validated measure of media literacy.
Try Rewriting A Headline
Here is an invented headline for practice: Every Teen Now Depends On AI. It is deliberately unsupported and is not a quotation from a publication. Before accepting it, ask what depends means, which teenagers were included, and whether the evidence measured dependence at all. Replace the headline with a sentence that states only what the original measure supports. Repeat the exercise with a real claim you encounter, retaining its source and date. Avoid turning your rewrite into the opposite exaggeration. Evidence that fails to establish a sweeping claim does not automatically prove there is no issue worth discussing. Your aim is to make the uncertainty visible enough that someone else can ask a better question.
Bring A Question Home, Not A Verdict
A population finding cannot establish how one particular child uses a tool. Clued’s suggested next step is to choose one ordinary activity to discuss: looking something up, practising a subject or making an image. Ask what the tool contributed and what remains uncertain. Keep that conversation separate from claims about all young people, and avoid interpreting a single answer as a diagnosis or a complete account of someone’s habits. For practical checking exercises, read Your Child Uses AI. Can They Explain When It Is Wrong?. For a different decision, explore Should AI Predictions Decide A Degree Choice?. Leave with one claim you can explain accurately and one question you still need to investigate.
Sources And Further Checks
Primary sources checked on 08-10-2026. Course and policy details can change; check the current official information before acting.
Ofcom: Research Release, 21 May 2026 Ofcom: Children And Parents Media Use And Attitudes Report 2025–6AI-assisted production. How We Produce Stories · Report A Correction
