Build log/Education

The Children’s code for edtech: what the ICO expects from learning apps

The ICO’s Age Appropriate Design Code sets 15 standards for online services likely to be used by children in the UK. Here is what each one means for an edtech product.

The short answer

  • The Age Appropriate Design Code, or Children’s code, is a statutory code of practice from the ICO for online services likely to be accessed by children under 18 in the UK.
  • It sets 15 standards, including high privacy by default, geolocation off by default, no nudge techniques that weaken privacy, and a data protection impact assessment.
  • Whether it applies to an edtech product sold to schools depends on who decides how pupils’ data is used, so take legal advice on your position.

The Age Appropriate Design Code, usually called the Children’s code, is a statutory code of practice from the Information Commissioner’s Office (ICO) that sets 15 standards for online services likely to be accessed by children in the UK. For an edtech product it means high privacy settings by default, collecting only what each feature needs, keeping location off and avoiding design that nudges children to share more. It is a practical guide and does not replace legal advice.

What the Children’s code is

The code was made under section 123 of the Data Protection Act 2018. It came into force in September 2020 with a 12-month transition period, so services have been expected to conform since September 2021. It is not a separate law. It explains how UK GDPR and the Data Protection Act 2018 apply to children’s data in online services, and the ICO and the courts must take it into account. A service that ignores it will find it hard to show it meets UK GDPR, and fines under UK GDPR can reach £17.5 million or 4% of annual worldwide turnover, whichever is higher.

Who the code applies to

The code applies to information society services likely to be accessed by children. An information society service is, broadly, an online service normally provided for payment, including one funded by advertising, that a user requests individually, such as an app, website, online game or connected device. “Children” means anyone under 18. “Likely to be accessed” means it is more probable than not that children will use the service, whether or not they are the intended audience.

For edtech, the position depends on how the product reaches children.

  • Sold to families or directly to learners, such as a home tutoring app or a language app for children: in scope.
  • Sold to schools, where the school decides how pupils’ data is used and you act only as the school’s processor: the code may not apply to you directly, though the school still has to meet UK GDPR and will expect your help.
  • Sold to schools, but you also use pupil data for your own purposes, such as product development, analytics or marketing: for that processing you are a controller, and the code is likely to apply.

Where the line falls in your contracts and data flows is a legal question. Take legal advice, and write down your reasoning so you can show it later. The Online Safety Act 2023, enforced by Ofcom, is a separate regime that may also apply if children can share content with each other in your app.

The 15 standards in plain English

StandardWhat it asksWhat it means in a learning app
1. Best interests of the childMake the child’s best interests a primary consideration in designWeigh the learning benefit of each feature against its use of data
2. Data protection impact assessmentsAssess and reduce the risks to children before processing their dataA DPIA before launch, updated for each feature that uses children’s data
3. Age appropriate applicationEstablish users’ ages with certainty that matches the risk, or apply the code to everyoneUse age bands from sign-up, or treat every user as a child
4. TransparencyPrivacy information that is concise, prominent and in language children understandShort explanations at the moment a feature asks for data, written for the age group
5. Detrimental use of dataDo not use children’s data in ways shown to harm their wellbeingTake care with streaks, rewards and notifications designed to keep children in the app
6. Policies and community standardsUphold your own published terms, policies and standardsIf you say chat is moderated, moderate it
7. Default settingsHigh privacy by default, unless there is a compelling reasonProfiles private, sharing off, analytics minimal
8. Data minimisationCollect and keep only the minimum data for the parts of the service a child actively usesSeparate choices for optional features, with no data collected for features left off
9. Data sharingDo not disclose children’s data without a compelling reasonNo sharing with advertisers or data brokers, and a list of every processor
10. GeolocationLocation off by default, with an obvious sign when it is onMost learning apps do not need location at all
11. Parental controlsExplain them to the child, and show an obvious sign when a parent is monitoringIf parents can see activity, the child can see that they can
12. ProfilingProfiling off by default, and only with measures to protect the childAdaptive learning can count as profiling, so record why it is needed in the DPIA
13. Nudge techniquesNo nudges that lead children to share unnecessary data or weaken their privacyEqual weight for “yes” and “no” on every privacy choice
14. Connected toys and devicesConnected toys and devices must include effective tools to meet the codeApplies to learning hardware such as tablets or smart speakers for children
15. Online toolsProminent, accessible tools for children to use their data rights and report concernsA visible “ask about my data” or “report a problem” route in the app

High privacy by default

Default settings matter because many users, and children in particular, keep whatever a service starts with. Under the code, every setting that affects privacy should begin at its most protective option, and the child or parent can relax it when there is a reason. In a learning app that typically means:

  • profiles visible only to the child, their parent and their teacher;
  • leaderboards that use nicknames, or are off;
  • chat with other learners off, or limited to a moderated class group;
  • analytics that collect only what is needed to run and improve the service;
  • geolocation off, and any setting that makes a child’s location visible to others switching back to off at the end of each session.

Nudge techniques and engagement features

A nudge technique is a design feature that steers a choice, such as a large bright “Accept” button next to a small grey “Not now” link. The code rules out nudges that lead children to share unnecessary data or to weaken or turn off their privacy protections. Nudges towards the more private option are acceptable.

Edtech has its own temptations here, such as streaks, badges and notifications that reward time spent in the app. They can support learning, but the detrimental use standard asks you to avoid features that keep children engaged beyond what is good for them. For each mechanic, ask whether it rewards learning or simply time in the app, and record the answer in your DPIA.

Parental controls in practice

Parents have a legitimate role, and parental controls are common in edtech. The code accepts that, but asks you to explain controls to the child in language suited to their age, and to show the child an obvious sign whenever a parent is monitoring their activity or location.

In CurioKids, the parent’s app is where lessons are booked, joined and followed, and homework happens in a child mode opened with the parent’s PIN. That split, with the parent as account holder and a separate, limited space for the child, is a useful pattern for products used by young children. We cover the wider product in building a tutoring platform, and how lessons are personalised in a lesson for the child.

Doing the DPIA

A data protection impact assessment, or DPIA, is a written assessment of how a service uses personal data, what could harm the people involved and how you will reduce those risks. The code expects one for any service in scope, done before launch and kept up to date. The steps:

  1. Map every piece of data you collect, where it goes and why.
  2. Decide how you will establish users’ ages, and record why that level of certainty is enough for the risks.
  3. For each feature, assess the risks to children of different ages and stages of development, including physical, emotional and developmental harm as well as data risks.
  4. Seek the views of children and parents where you can, and of your data protection officer if you have one.
  5. Record the measures you take for each standard, and the reason for any departure from a high privacy default.
  6. Review the DPIA with each significant release.
Designing a product for children? We plan the Children’s code standards into the first design. See our education work.

Questions people ask

Does the Children’s code apply to my edtech app?

It applies if your app is an online service likely to be accessed by under-18s in the UK. Apps sold to families or directly to learners are in scope. If you sell only to schools and act purely as the school’s processor, the code may not apply to you directly, so take legal advice on your position.

Is the Children’s code a law?

It is a statutory code of practice made under the Data Protection Act 2018, and it does not create new legal duties of its own. The ICO and the courts must take it into account when deciding whether a service meets UK GDPR. Ignoring it makes compliance very hard to show.

Does an edtech app need a DPIA?

Yes, if it is in scope of the Children’s code, which expects a DPIA before launch. Offering online services directly to children is also on the ICO’s list of processing likely to need a DPIA under UK GDPR. Update it whenever a new feature uses children’s data.

Written by Flavoureak

We build and run Estimark, CurioKids and CalcStack, and we build web and mobile software for other businesses. Start a project.