Edtech Safety Certification: What England’s Plan Means for Your School

The Department for Education (DfE) has proposed independent edtech safety certification for filters, data-heavy apps, pupil-facing artificial intelligence (AI) and personalised learning tools. Here is what a badge would cover, what it would not, and what schools can do in the meantime.

Last week the Department for Education (DfE) opened a consultation on edtech safety certification, and it could change how every school in England chooses its apps. The idea is simple enough to explain in a staff briefing. Higher-risk products, including artificial intelligence (AI) tools that pupils use directly, would be checked by independent assessors against safety criteria, then listed on a public register with a badge. Most school leaders I talk to will welcome that. The catch is in the small print. Nothing would be certified for at least 18 months, and the badge would say nothing about whether a tool helps pupils learn. Meanwhile, the AI tools teachers build for themselves would be left out entirely.

The short version: the DfE wants independent edtech safety certification for four kinds of product: web filtering, tools that do high-risk processing of children’s data, pupil-facing generative AI, and personalised learning apps. Certification would cover safety, not effectiveness, and would not be compulsory. The consultation closes on 4 February 2027. Until a scheme exists, the checking is still your job.

An open steel padlock standing on a dark laptop beside its keyboard, softly lit against a bright blurred background
An open padlock on a laptop. Photo by Yuri Samoilov (yuri.samoilov.online), via Wikimedia Commons, Creative Commons Attribution 2.0, cropped.

What the DfE is proposing

The consultation was published on 10 November 2026. The full consultation document runs to more than 50 pages and is worth an hour of any digital lead’s time. It names three risks it wants to tackle: unlawful use of children’s personal data, exposure to harmful content, and harm to children’s cognitive, social and emotional development and mental health.

The model borrows from product safety in other industries. The United Kingdom Accreditation Service would accredit independent assessment bodies. Suppliers would pay one of those bodies to test their product. Products that pass would carry a badge and appear on a register run by the DfE, showing what was assessed, who certified it and when the certificate expires. Generative AI products might have to be recertified every year because they change so quickly. Other products would face full recertification every three years, with lighter annual checks in between.

The DfE says schools asked for this. In its Technology in Schools Survey 2025, half of school leaders said safeguarding and data concerns were holding back their use of technology. Anyone who has tried to assess a new app on a Tuesday afternoon, with a supplier’s glossy privacy page and no technical expertise, will recognise the problem.

Which products edtech safety certification would cover

The scheme is deliberately narrow. It covers four categories, and a product can fall into more than one.

The four categories

Filtering products are the first, assessed against the existing filtering and monitoring standards. The DfE is also asking whether certified filters should have to cope with real-time and AI-generated content. The second is any product doing high-risk processing of children’s data, such as management information systems, safeguarding and behaviour platforms, wellbeing apps, AI tutors and platforms supporting pupils with special educational needs and disabilities. Third come pupil-facing generative AI tools, including chatbots, tutors and content generators. Last are pupil-facing personalised learning tools, which adjust content, pace or difficulty for each child.

For the generative AI category, products would be judged against the eight areas in the DfE’s existing generative AI product safety standards. These include filtering, logging of prompts and responses, protection against emotional dependence, responding safely in a mental health emergency, avoiding manipulative design and reducing the risk of what the document calls “cognitive deskilling”. That last phrase will ring bells for anyone who read our piece on cognitive offloading and AI.

For personalised learning apps, the DfE expects the focus to be design features that add unnecessary pressure, gamification, or features that keep children engaged without a teaching reason. One consultation question asks directly about rewards, streaks, leaderboards and notifications. Think of the maths app where a Year 7 pupil is nagged at 9pm to protect a 40-day streak. Under this scheme, that design choice becomes a safety question.

What is left out

Office tools such as email and word processors are excluded, along with static content such as digital textbooks, hardware, and staff-facing tools that do not do high-risk processing of children’s data. Monitoring products are left out at launch, because their safety depends so much on how staff act on alerts.

The exclusion I would underline in red pen is this one. Teacher, college or school-created AI agents, custom tools and wrappers used only within that setting would be exempt, and the school would stay responsible for making sure they are safe and lawful. So the commercial revision chatbot your trust buys would need a badge, while the homemade revision assistant a keen science teacher builds for Year 11 over half-term would not. Pupils will meet both in the same week.

A safety badge will tell you a product has passed a test. It will not tell you whether your pupils learn anything from it, or whether the tool your own staff built is safe.

What a badge will not tell you

The DfE is clear that certification would assess safety, not overall quality or educational effectiveness, because effectiveness depends on how and where a product is used. It points schools to the Edtech Testbeds programme with the Education Endowment Foundation for that question instead. I think that is the honest position, but staff and parents will not necessarily read a badge that way. A certified AI tutor could be perfectly safe and still make only a modest difference to learning, which is roughly what the trials in our article on AI tutoring found.

Timing matters too. Even if the consultation goes smoothly, the document says at least 18 months would be needed to set up the scheme before any product could be certified, followed by a transition period. The government response is due in spring 2027. On that timetable, I would not expect the first badges before 2028, and pupils are using these tools now.

Then there is the question of how strong the expectation would be. The DfE proposes adding it to Keeping Children Safe in Education (KCSIE), the statutory safeguarding guidance. Schools buying or using an in-scope product would be expected to get “proportionate assurance” that it meets DfE safety standards. Certification would normally provide that, but schools could rely on other credible evidence if they explain why it is equivalent. The DfE has no power at present to make certification compulsory, and the consultation asks whether legislation might be needed later.

Mixed messages on chatbots

On 16 November, the early education minister Paul Waugh was asked in the Commons about AI chatbots in schools. As Tes reported, he replied: “It’s not in our schools, because our guidance is very clear that AI chatbots shouldn’t be used in schools.” Tes noted that this appeared to go further than current guidance. The consultation itself says pupil-facing AI tools “have the potential to transform education by providing tailored support and learning opportunities to every child”.

Both cannot be the working rule. If the policy were no chatbots in schools, there would be little point designing a certification category for them. My reading is that the written proposal is the better guide, but expect governors and parents to quote the minister at you.

What schools can do before any badge exists

None of this needs to wait for 2028. Here is where I would start this term.

  • List every tool pupils use and sort it into the four categories. You may find more pupil-facing AI than you expected, sometimes switched on by default inside products you already pay for.
  • Use the generative AI product safety standards as a supplier questionnaire now. Ask which standards the product meets, how prompts are logged and who sees alerts.
  • Look hard at engagement features in personalised apps. If a streak or leaderboard exists to keep pupils clicking rather than learning, say so to the supplier.
  • Put homemade AI tools through the same approval route as bought ones, with the designated safeguarding lead and data protection officer involved. Our checklist on AI and data protection covers the questions to ask.
  • Respond to the consultation. Many questions are aimed at schools, and the DfE admits the evidence on design features such as gamification is still developing.

I am glad the DfE is doing this. Schools should never have had to judge the safety of complex AI products alone. But a register of safe products is a floor, not a strategy, and the hardest questions about which tools help pupils think, and which do the thinking for them, will still land on teachers’ desks.

Frequently asked questions

What is the Department for Education’s edtech safety certification scheme?

It is a proposal from the Department for Education, published for consultation on 10 November 2026. Independent assessors would check higher-risk education technology against safety criteria, and products that pass would get a badge and a place on a public register. It would apply to settings in England.

Which products would need edtech safety certification?

Four categories: web filtering products, tools doing high-risk processing of children’s data, generative artificial intelligence (AI) tools that pupils use directly, and personalised learning apps for pupils. Office software, static content, hardware and most staff-only tools are excluded, as are AI tools a school builds for its own use.

Will schools have to use certified products?

Not at first. The proposal would add an expectation to Keeping Children Safe in Education (KCSIE) that schools get proportionate assurance about in-scope products. Certification would normally provide it, but schools could use other credible evidence if they explain why it is equivalent.

When could the first products be certified?

Not soon. The consultation closes on 4 February 2027 and the government response is due in spring 2027. The DfE says it would need at least 18 months to set up the scheme before any product could be certified, followed by a transition period.

Would an edtech safety badge change what your school buys, or do you already have your own checks in place? Tell us how you vet new apps and AI tools in the comments.

Clara Whitfield
Clara Whitfield

Clara Whitfield is the pen name used for editorial articles published by Technology for Learners. These articles cover artificial intelligence (AI) and digital tools in teaching and learning, and cite the research, guidance and reports they draw on so readers can check the evidence.

Articles: 33
Verified by MonsterInsights