We’ve packed lots in this week as we’ll be taking a summer break from now until the new term… Buckle up and see you in September!

Congratulations to the 27 teenagers who were the first to graduate from the London AI Campus’s two-year enrichment programme, AI Connect.

We caught up with London AI Campus Director Laura Flynn-Coley on the campus in King’s Cross earlier this year to find out more about this unique initiative.

AI Connect launched in November 2024 and takes up to 30 students across Years 12 and 13 from secondary schools across London. Entry is competitive and students must be studying A-Level Maths. Laura explained that the programme, which the students tackle outside their regular school hours, moves through distinct phases:

Fundamentals of AI – understanding the ‘black box’: data, algorithms, generative output and predictions.

Robotics – delivered with an external partner, it builds from basic electronics and logic through to robots that can “hear, touch and see,” culminating in a student-led showcase.

AI and sustainability – students research a topic in teams, supported by mentors from organisations such as UCL and the Alan Turing Institute, then present their projects to a panel followed by audience questions.

The London AI Campus has two other strands: the AI Opportunity Centre, and teacher CPD. The AI Opportunity Centre is the more scalable strand, open from Key Stage 3 to Key Stage 5, offering broader enrichment across subjects rather than a fixed curriculum – examples include AI and democracy, an escape room, and sessions on AI and art or music.

Female participation in the programme is unusually high – and that’s no accident. For Laura, a focus on gender representation is non-negotiable. As she told us, “I’ve taught computer science for 17 years, and the disparity with male to female representation is absolutely horrendous.” As a result, as well as an emphasis on visible female role models, the curriculum is designed with gender inclusivity in mind.

Key pedagogical principles include:

Student choice over prescription: students choose their own projects and roles rather than being assigned rigid tasks, since prescriptive curricula are cited as a particular barrier to girls’ engagement.

Critical thinking built in throughout: students are pushed not just to explore what AI can do, but its drawbacks and ethical implications.

Authentic accountability: presentations to real panels and industry experts, including high-profile settings like Google’s headquarters, rather than performative classroom exercises.

Mentorship: each student is paired with an industry mentor (many from Google, DeepMind, GSK and AstraZeneca) whom they meet roughly every six weeks – something Laura frames as meeting a basic psychological need to feel valued and supported.

Learning through failure: the programme aims to build resilience by giving students structured project briefs with room to fail and reflect, which Laura contrasts with a school culture that rarely allows this.

The through-line across both the gender and pedagogy angles is the same: creativity, choice and real-world stakes are what get students – particularly girls – to engage meaningfully with AI, rather than a curriculum that tells them exactly what to do.

Of course, 30 students is a drop in the ocean, and the need and demand for this kind of training is huge. What’s the answer? For Laura, curriculum reform and much more teacher support are crucial if pupils are to get the grounding in computing and digital literacies they need in the AI age.

“The current A-Level computing curriculum is so out of date, it’s shocking,” says Laura. She advocates for an enhanced level of creativity and project development within the computing curriculum, a greater focus on digital literacy at Key Stage 3, and a strong emphasis on students understanding logic and problem-solving.

She also points to the severe shortage of computer science teachers, compounded by school leaders and policymakers not giving staff the necessary time and recognition to master emerging AI skills and develop new curricula. After all, she says, “if you don’t have happy teachers, you don’t have happy students.”

AI roundup

Shuffling around
UK readers may have noticed a new PM and government this week. Among the shuffling, the MInister for AI, Kanishka Narayan, now sits in cabinet, DSIT has gone and DCMS has acquired ‘Digital’ in its name. The new Office for PM and Cabinet includes an AI Taskforce, a PM adviser on AI, and overall government strategy on AI while the Cabinet Office takes lead on AI adoption across the civil service and AI strategy, public sector AI adoption and the AI Security Institute.

And, in education, Schools Week has 10 facts about the new education secretary Lucy Powell, including that she’s from a family of teachers and when she was shadow minister for education in 2015 she called for free schools and academies to be brought under the oversight of local authorities.

But where are the otters?
Ethan Mollick has published one of his regular posts about which AI to use to do stuff, and, sadly, seems to have abandoned his much-loved otter on a laptop on a plane test in favour of a brutalist city building game. Oh well. He points out that while the AI landscape might seem to have got more complicated with the agentic system, using AI is also getting easier because the AI is increasingly working out how to solve problems without you knowing the details. Plus, prompting skills are becoming less important than simply instructing the AI by asking for what you want and correcting the AI when it doesn’t get your intentions. But his advice remains the same - pick ChatGPT or Claude, give the AI a task you need doing, see how it does and ask it to do better.

Stop being so nice and helpful!
In this insightful two-part blog series on creating an AI learning coach, LSE academic Savvas Verdis describes how his biggest challenge was to teach the AI model to go against its own trained instincts and not reframe a student’s vague answer into a smarter one. His prompt file (which he has published) instead includes a set of rules such as never both praise and re-explain in the same turn, diagnose a flawed example before drafting your own, and push back on overconfidence.

Science fiction that happened
Simon Willison has a good rundown of OpenAI’s accidental cyberattack against Hugging Face (in which OpenAI says its advanced AI model went rogue). He points out that this is exactly the sort of thing one should expect these models to do – “It turns out relentless proactivity is the defining trait of this new generation of Mythos-class models. If you set them a goal and give them a way to get there, even inadvertently, they will figure it out” – and warns against dismissing the story as a marketing trick. In another account of the break out, Transformer says,

“AI safety researchers have long hoped that we would receive a “warning shot” of dangerous misalignment behaviour before too much damage was done. An AI system breaking out of its testing environment and hacking into another company’s infrastructure in order to steal the answers to its test is about as clear a warning shot as you could get.”

Gemini Notebook
NotebookLM is now Gemini Notebook and Steven Johnson explains how it got such an odd name in the first place (along with thoughts on philosophy, memory etc)

Teenagers, pornography and explicit deep fakes in the age of AI
The latest research publication from Common Sense Media shows almost half of teenagers have already seen AI-generated sexual material, and almost a quarter have seen it depicting themselves or someone they know. Nearly one in five (boys more than girls) said they had made AI sexual content themselves or knew someone who had. Among that group, nearly two-thirds said the images involved people inside their own social world: a friend, an ex, or a classmate. The report provides advice for parents and ways to start conversations with their children.

• Common Sense Media’s Youth AI Institute has a new AI product evaluation risk assessment on Google AI search.

The overlooked upside of AI in schools for school improvement
AI is giving teachers a new way to shape and sharpen their ideas for school improvement and that’s something to welcome, argues school head Adam McRoy in the TES.

AI and accessibility
Excellent episode of Jisc’s Beyond the Technology podcast discussing how AI is improving accessibility across higher education, including AI-powered captioning, student-led innovation, digital inclusion and the future of accessible learning. Jisc also has a roundup of conversations with its assistive technology community, with more on providing live captions.

Same brains. New engineered traps
Just as the ideas of ‘learning styles’ and of ‘digital natives’ have been discredited, so too should the notion steadily gaining traction that Gen Alpha needs a different kind of learning with shorter snippets of information, argues Christian Turton. Rather than feeding into the harms of passive scrolling and decreasing attention spans, we should be focusing on helping young people to develop critical thinking, self regulation and metacognition.

Quick links

• We enjoyed this cartoon by Stephen Collins, Fun Summer Holiday Ideas for Arguing About Screen Time, as we anticipate the next six weeks with the children in our lives.

• The latest Online Safety Network newsletter is out.

• Wondering about falling rolls in your school? FT journalist John Burn-Murdoch says that while financial instability and housing shortages act as major barriers to people starting families, the effect of smartphones is part of the picture for birth rates declining everywhere. See also, How AI sexbots could destroy the world.

• Psychologist Candice Odgers questions the dominant narrative that the best way to address online harms is a social media ban for teenagers and says that the current debate around smartphones obscures some of the biggest issues facing teenagers. She’s also challenged Jonathan Haidt’s interpretation of research findings.

• Do Meta’s smartglasses pose a safeguarding risk to children?

We’re reading, listening, watching…

Play doesn’t stop when the screen goes dark
According to this fascinating blog post from the LSE’s Parenting for a Digital Future, 70% of South Korean infants and toddlers begin using smart devices before the age of three, and the average age of a child’s first personal smartphone is 3.6 years. By the time children are between three and nine, they are spending more than three hours a day with media. YouTube, Netflix and games like Minecraft and Roblox are widely used even among preschoolers. Yet, contrary to the usual narrative on screentime, this study finds “their digital lives were stitched together with play, care and culture” as the researchers watched “digital prompts flow straight into hands-on, social activities. Screens were catalysts – not endpoints – for making, moving, and telling stories together.” It’s blended and spontaneous.

Catching them young
Meanwhile, in China, the race is on for AI-savvy teenagers as Chinese companies steadily expand their pipeline by recruiting ever younger talent long before it reaches the workforce.

Ceci n’est pas une pipe
The Wind and the Song is a short film from Fully Human, a PSHE Association initiative. As chatbot use among children and young people explodes, the film demonstrates that now is the time to confront urgent questions about what it means to be human alongside technology that pretends to be us. Animator, director and graphic artist Lewis Heriz uses old techniques such as stop motion and found footage, alongside AI, to examine the consequences of humanised technology and the peculiar paradox of our time; that to access reality we must almost constantly engage with illusions.

The age of reading is over?
We were about to share this audio version of Rose Horowitch‘s long form article from the August issue of The Atlantic about the decline in reading, how digital media is hijacking attention and how short form video is contributing to a post-literate culture. And then we heard ‘the fact that you are reading this article means that you are part of a subculture devoted to text – and we appreciated the irony... So, how about joining Naomi Alderman’s couch to 5K for reading instead? Combine with the English and Media Centre’s summer reads for secondary students and teachers.

Tech mindsets for teachers
A plug for more summer holiday reading! Our former Connected Learning Centre colleague Krzysztof Jurek has just published a book, The Teacher’s Tech Mindset - A Practical Guide to Building Confidence and Clarity with School Technology. Krzys draws on many years’ experience designing, building and maintaining schools’ computer networks and providing technical support to schools. His book is less about becoming a tech expert than about developing a proactive approach to troubleshooting and communicating, building the understanding, habits and mindset that help people feel more comfortable and capable when working with technology. The book is illustrated by another of our CLC colleagues, Blossom Carrasco. Well done CLC-ers!

Give it a try

Ghost Font
Ghost font is an anti-AI font that can be read by humans but not leading AI models. Type your text, then download and share the video clip containing your message.

• Your suggestions of useful, intriguing, quirky digital things to try are always welcome - drop us a line on email or catch us on LinkedIn.

Connected Learning is by Sarah Horrocks and Michelle Pauli

We’re zipping straight to the news and views this week but don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

Instagram AI image reversal
Last week, in our post about AI images and safeguarding, we mentioned (via Wired) that, as part of Meta’s Muse Image model rollout, anyone will be able to use Instagram photos in AI images unless the poster has specifically opted out. This has now been reversed. TechCrunch reports that Meta has conceded that the new feature has “missed the mark” and will not be rolled out.

More on AI images
A Teacher Tapp survey suggests more than one in four secondary teachers say a pupil at their school has created an AI-generated sexual image of a classmate, reports Schools Week.

Hello AI
The new (free!) issue of Hello World magazine focuses on critical thinking in the age of AI. Teachers share how they are helping students to think beyond AI answers to develop critical-thinking skills while learning to use AI thoughtfully in the classroom. There’s an accompanying podcast discussing the difference between using cognitive load to support learning and using AI in ways that replace the thinking process entirely.

You say tomato…
Intriguing suggestion that Anthropic’s Claude behaves differently depending on which language it is using. A study of sycophancy in the model found that it was most deferential, briefer and warmest in Arabic and more cautious, rigorous and depth seeking in English. It is most candid about its own flaws in Dutch. But does this produce inconsistent moral reasoning or priorities? Still to be explored.

Very powerful. Very useful. Not magic.
Handy glossary of 39 AI terms and concepts, from LLM and prompt to agentic loop, retrieval-augmented generation and context window.

Something! Shriek!
No parent or teacher of 7-11 year olds will be unaware of the comic-book phenomenon that is Bunny Vs Monkey. Now creator Jamie Smart has pleaded with teachers not to use his characters in AI-created resources or literacy schemes, having shared one particularly egregious example (now deleted).

Quick links

• The EU’s Special Panel on Child Safety Online published its report, Protecting and Empowering Minors in a Digital World, this week. It’s intended to inform future actions by the European Commission and EU Member States to reinforce child safety online.

• The UK government has updated its data protection in schools guidance, including sections on procuring edtech.

• Social media curfews are the talk of the town after the UK government announced default overnight curfews from midnight to 6am for 16 and 17 year olds. The BBC has more detail and the Today in Focus podcast simply asks if it’s pointless.

• The EU is also planning to impose a minimum age for young internet users to access social media without parental supervision.

  

• Meanwhile, a UCL report suggests that school smartphone bans are overly simplistic and (unsurprisingly) not supported by young people.

• And what about their parents? A research report finds that parents and caregivers’ attachment to screens can exacerbate anxiety and insecurity in children and teens. And, in Uruguay, the rollout of high-speed internet worsened early childhood development – not because children spent too much time in front of screens, but because their parents did.

• TikTok faces an Ofcom investigation over child age checks. The regulator has looked into how the platform uses technology known as “age inference”. This relies on estimating how old a user is based on how they use the platform, such as the videos they watch or others they interact with, reports the BBC. Kate Davies from Ofcom said she had “serious doubts” over whether such tools are good enough at checking the age of users.

• NEU general secretary Daniel Kebede reflects in the TES on how new UK PM Andy Burnham can take on social media legislation and ensure it truly delivers for schools.

We’re reading, listening, watching…

Parents put your phones down
We’ve previously featured a radio programme about Baroness Kidron’s new book, Users. In this powerful Intelligence Squared podcast interview, she talks in detail about the unchecked power of Big Tech: the technologies designed to capture our attention, which can quickly expose children to harmful content, and the compulsive design, data harvesting and surveillance that underpin it. Kidron also makes the plea for parents to put their phones down and give children their attention.

Consuming AI ethically

The whole July/August issue of Ethical Consumer Magazine is devoted to AI. Articles include guides to AI tools, social media, web browsers and investigations of privacy, military use, data centres, climate, tax and corporate power.

How AI is changing language
The interplay between humans and AI is creating a kind of linguistic hall of mirrors, argues David Shariatmadari in this fascinating Guardian delve (a word that gets a fair bit of attention) into how AI is changing language but will never achieve the literary innovation that humans, as social animals, can with our “human flesh, with its spikes of adrenaline, rushes of dopamine, craving for social contact, all of which find expression in language’s structure and the way we use it.”

Give it a try

PlotLines

Oh, we do like delightful geekiness like this. In PlotLines someone (we’d love to say who but can’t find an ‘about us’ page on the site - please let us know if you know) has mapped every journey in classic books (think Pride and Prejudice, Anna Karenina, Ulysses, Madame Bovary etc) and plotted them on a hand-tinted period map, using the real 1890s Ordnance Survey. Explore the journeys book by book or see them all plotted out on the atlas.

Connected Learning is by Sarah Horrocks and Michelle Pauli

The updated Keeping Children Safe in Education (KCSIE) 2026 was released this week and there are some important changes. KCSIE is statutory guidance issued by the UK Department for Education outlining the legal duties schools and colleges in England must follow to safeguard and promote the welfare of children under 18.

In the updates, as well as referring leaders to the resources Michelle worked on with Chiltern Learning Trust to help school and college staff use AI safely and effectively, there are some specific actions schools need to take before KCSIE 2026 comes into force on 1 September.

Dan Fitzpatrick has written a helpful post listing eight of these:

1. Update your child protection policy to name AI-generated imagery. KCSIE now explicitly requires policies to cover nudes and semi-nudes “including those generated using AI”. If your policy doesn’t say this, it’s out of date on day one of term.

2. Brief your DSL on AI-generated child sexual abuse material (CSAM).

3. Add AI chatbots to your online risk assessment.

4. Get your governors reading the DfE’s Generative AI guidance.

5. Check every AI tool against the DfE’s product safety expectations.

6. Do your filtering and monitoring review properly.

7. Plan deepfake education into the curriculum.

8. Retrain staff on the new definitions. AI-generated images are now folded into the definition of nudes and semi-nudes throughout the document. Every member of staff needs to treat a deepfake incident with the same seriousness as a real image.

We’d like to pick up on a specific issue affecting schools around AI-generated imagery. There’s growing public awareness of the issue with the National Crime Agency now recommending that parents should not put photos of their children on public display online. The Internet Watch Foundation has also highlighted the case of an unnamed UK secondary school which was subjected to a blackmail attempt after criminals used the institution’s website or social media accounts to take the photos of schoolchildren and then, using AI tools, turned them into child sexual abuse material. The blackmailers sent the images to the school and threatened to publish them online if they did not receive money.

The issue is now no longer simply whether organisations have permission to use photographs, but whether it is still proportionate to publish identifiable children’s images publicly when those images can be scraped, copied and manipulated in new ways. With parents quite reasonably asking about the steps being taken to protect their children’s images from misuse, in practice, many organisations are being pushed between a small number of unsatisfactory options: carry on as before and accept the risk, stop showing images of children altogether, or send images off site to a third party for processing. Are we about to see images of empty classrooms and playgrounds on school websites?

Against this background, new tools such as Anonemo are emerging as possible alternative options: a way to keep the real moment, setting and story visible while reducing the exposure of real children’s faces online, using local-only processing on the school’s own hardware to anonymise the faces while preserving the moment. Expect to see more workarounds like this emerge as the implications of KCSIE sink in across the sector.

AI roundup

Instagram image risks
Continuing the theme of AI, images and safeguarding, any educators using Instagram should note that, as part of Meta’s Muse Image model rollout, anyone will be able to use your Instagram pictures in AI images unless you specifically opt out. As Wired explains, “public Instagram profiles are now automatically opted into being fodder for generative AI remixes. All someone has to do is tag your account’s profile in a prompt – if it’s public – and they can use Meta AI to generate an image using your likeness.” Wired also details how to dig into the app’s settings to turn off this ‘feature’.

Making greener choices
Clear five tips from Dutch tech ethics specialist Hidde de Vries about AI and sustainability and how to make greener choices when using AI, from choosing smaller and specific models to recognising greenwash. Thanks to Matt Jukes for sharing.

Ofqual chief warns against AI coursework
Plans for written coursework in qualifications for 16 to 18 year olds will face much more scrutiny according to Ofqual chief Ian Bauckham. According to Schools Week, he said, “For the qualifications to be worth more than the paper they are written on, we cannot normalise the idea that AI-generated output is a substitute for genuine human endeavour.” He cautioned against wider use of tech in the classroom but acknowledged the potential benefits of AI marking.

AI tutors ‘make more work’ for teachers
With the government planning to introduce AI tutors at the end of next year, experts have been giving views to MPs on the House of Commons Education Select Committee. Professor Neil Selwyn, of Monash University in Australia, shared his concerns that the amount of behind-the-scenes work needed to use the technology effectively risks “burdening teachers with quite low-level work” reports the TES rather than freeing them up to be pedagogy experts. Professor Rose Luckin told MPs, “We need to be really careful how we design and implement any AI tutor that we roll out. The best designed tool will not have the impact we want unless it is carefully implemented.”

Quick links

• A joint statement has been signed by 48 of the Online Safety Act Network partners outlining the opportunities for online safety which lie ahead for the new UK prime minister. Pair with Baroness Kidron’s opinion piece, which suggests the tech lobby has thwarted attempts to hold the sector to account and asks the UK government to commit to several measures, including that any tech deployed in the UK will respect the privacy, rights and safety of children, and a commitment to use the data of the BBC, NHS, UK innovators, creators and businesses to benefit the UK.

• The Digital Literacies Network has published picture books in PDF form that aim to bring into focus vocabulary associated with social media practices.

• Raspberry Pi has produced two new Pedagogy Quick Reads focused on Mayer’s Principles of Multimedia Learning, with examples for how to apply the principles in computing lessons.

We’re reading, listening, watching…

Bridging the digital divide
This long read from Card Catalog examines who can navigate information systems and who can’t, looking at the four layers of digital inequality (physical access, skills and literacy, quality of access and meaningful use) and the impact of AI. It also offers suggestions of what communities governments can do to close the gap.

Teenage kicks
A story from 2021 about a Dutch teenage hacker which is even more relevant today in relation to young people’s vulnerability online. Edwin Robbe had a troubled life, but found excitement and purpose by joining an audacious community of hackers. Then the real world caught up with his online activities. Pair with this week’s excellent London Centric’s story of a teenage millionaire hacker from Tower Hamlets.

LLMs as material: made of everyone

In her newsletter Linda Liukas reflects on her recent visit to Reggio Emilia, on Loris Malaguzzi’s theory of the hundred languages of children (the conviction that a child knows and speaks the world in a hundred ways, and that clay, wire, paint, shadow, and song are each a language) and how this relates to the materiality of large language models:

• Language models are a material made of everyone and everything.

• Language models are a material made of probabilities.

• Language models are a material that reads in pieces and thinks in places.

• Language models are a material that leans toward you.

Give it a try

It’s just too hot this week! Your suggestions of useful, intriguing, quirky digital things to try are always welcome - drop us a line on email or catch us on LinkedIn.

Connected Learning is by Sarah Horrocks and Michelle Pauli

We’re zipping straight to the news and views this week but don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

Google’s AI push in schools
Dan Fitzpatrick has written an article for Forbes and made a podcast episode about Google’s latest, and extensive, AI for education updates, which are building up the Gemini ecosystem across Classroom, NotebookLM and Chromebooks. He asks questions about their true impact:

“Google Classroom is rapidly moving from a place where work is just set by the teacher and handed back in by the student, into a place where AI-assisted learning happens on the platform. Work designed, delivered and monitored all within Google Classroom, with the support of Gemini. Strip the announcement back and it collapses into three things. Context. Control. Insights.”

Excellent AI CPD
Join Christian Turton, edtech lead at Chiltern Learning Trust and DfE project lead, for a free, interactive cpd session on the safe and effective use of AI in education, exploring the new DfE training materials, key themes and links to KCSIE. It’ll be on Zoom on 1 September, 1.45pm - 3.00pm. Sign up here

A layered approach to AI and safeguarding
Jisc’s Michael Webb has written a really good, detailed examination of AI and safeguarding and how it all plays out at a technical level. It’s aimed at colleges but equally useful for schools.

Code.org becomes CodeAI
Code.org has rebranded as CodeAI, dedicated to the vision that every student in every school has the opportunity to learn about AI and computer science as part of their core K-12 education.

Spot when AI is fooling you
Sam ‘Slow AI’ Illingworth is running a free webinar on 22 July on How to spot when AI is following you. Find out when to use AI – and when to leave it well alone.

Stopping AI child sexual abuse imagery
The Internet Watch Foundation and National Crime Agency have published a guide for parents, carers and professionals on how to protect children’s photos and talk to children about the risks.

Quick links

• Research on the effect of screentime on babies within the 1,001 critical days links it to long-term negative effects on health and quality of life, with parents lacking guidance.

• The ICO’s audit of edtech companies whose products are widely used across UK primary and secondary schools finds systemic compliance gaps when it comes to children’s data.

• There’s a ‘keyboard gap’ affecting the chances of pupils who only have access to a smartphone rather than a laptop, says the Digital Poverty Alliance’s 2026 Schools Census Report.

• Norway bans generative AI use by pupils in schools.

• More on the Australian social media ban: a blog post on how dissenting academic voices need to change tack; new research on how Australian teens impacted by the social media ban are getting less news; and views on the National Media Literacy Strategy.

• Young people see school smartphone bans as punitive and counterproductive, creating negative displacement effects, according to new research from UCL.

• Nesta has tested NOMO (short for ‘No Missing Out’), a free app designed to help young people cut back on social media together, with 20,000 pupils, with interesting results.

• What does Google’s ‘teacher approved’ badge actually mean?

• In this webinar recording about embedding news literacy and civic engagement in primary education, Dr Gianfranco Polizzi explores how news and media literacy can be effectively integrated into primary school education.

We’re reading, listening, watching…

The human power of Bluey
Rose Luckin uses Bluey as a starting point for investigating the value of narrative, namechecking Jerome Bruner, and how the deepest forms of human experience come from our lived experience. And so, she argues, “the greatest risk posed by AI is not that machines become more capable than people. It is that we mistake fluency for understanding and begin to undervalue the forms of intelligence that are hardest to measure precisely because they have become part of who we are.”

Strict schools: taking discipline too far?
This BBC Radio 4 File on 4 programme isn’t directly technology related, but since we write a lot about children and young people’s mental health and wellbeing in this newsletter, we think it’s worth sharing. The File on 4 team investigates concerns about how some strict discipline measures are being implemented in schools across England causing significant harm to students.

Give it a try

WindowSwap
Can’t wait for the holidays to start? Pretend you’re somewhere else already with a random view out of a stranger’s window anywhere in the world. From Kolkata to Glasgow, New Jersey to Johannesburg, catch a glimpse of a different life.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Last week Sarah ran a workshop on digital technology at the ‘Inclusion Matters: Learning From What Works’ SEND and inclusion conference at the University of Gloucestershire.

The conference was led by Helen Williams and the day opened by Dr Tristan Middleton, Senior Lecturer in Inclusive Education at the University of Gloucestershire. Drawing on his experience as a teacher, SENCo and school leader, his keynote set the tone for the day by asking what ‘special’ really means in inclusive practice and what teachers and school leaders can do to make a genuine difference for individual learners through relational practice. Tristan’s keynote also put SEND in its sociopolitical and historical context, tracing some of our current challenges to policy decisions, beliefs and ideologies.

Sarah’s session covered several digital technology themes, and drew on the SEND network CPD programme that she worked on with colleagues in Camden and sharing of practice in Gloucestershire schools. Discussions covered:

• How digital literacy is an inclusion, equity and entitlement issue for all children and young people. For pupils with SEND, this entitlement is even more pressing. Where disability, special educational needs or other disadvantages already create barriers, low digital skills, limited access to devices, lack of confidence with technology and low digital and media literacy can compound existing disadvantage

• The overlooked power of accessibility features already built into devices and platforms, whether it is text-to-speech software or specific hardware settings. SENCos agreed that teachers and TAs need more training and the policy support to use these tools effectively.

• How important it is for teachers to understand Universal Design for Learning (UDL) principles as part of inclusive teaching when designing slides and other digital materials. UDL is about more than accessibility for students with SEND; it’s about creating flexible, diverse learning pathways that foster autonomy, reduce learner frustration and build a truly inclusive environment for every student in the classroom.

• Quality assurance for interventions: schools often lack a systematic approach to evaluating online interventions. We need to move beyond off-the-shelf solutions and look for evidence-based practice.

• Generative AI offers potential to support inclusion through tasks such as summarising, chunking information, creating scaffolds such as writing frames, knowledge organisers, sentence starters, translation – live or prepared, differentiation through reading age, dual coding through visual and audio materials, audio to text recording and summarising, audio descriptions of images and dealing with (AI generated) parent communication. However, innovation must be balanced with rigorous data governance.

A wide choice of workshops through the day spanned a range of SEND themes. Emotional wellbeing and pupils’ sense of self featured prominently. Tom Hoskins, founder of Chance is Change, shared his work supporting children’s belonging and agency through his new publication Welcome to Feelings Town. His Life Coach bus was also available for delegates to visit and explore throughout the day, giving a hands-on flavour to his work with individual pupils.

AI roundup

Guide to the environmental impact of AI
Jisc and the Environmental Association for Universities and Colleges (EAUC) have launched a practical guide to help universities and colleges address the environmental impact of AI in post-16 education and research. The core issue is that AI adoption is outpacing institutions’ ability to understand its environmental consequences, with sustainability considerations often missing from AI policy conversations. Key challenges include lack of transparency from technology providers about energy and carbon outputs, plus electronic waste from hardware manufacturing and disposal. But AI can also support sustainability goals when used thoughtfully – optimising campus energy use, supporting climate research, automating carbon accounting, and extending equipment lifespans through predictive maintenance. The guide recommends institutions audit their AI use, ask suppliers the right questions about environmental impact, integrate environmental literacy into staff training, and strengthen governance around AI’s sustainability implications. Jisc and EAUC are also offering a masterclass on 14 July and providing practical tools to help institutions embed sustainability into procurement and decision-making.

Updated European Commission OECD AI literacy framework
The OECD and European Commission have published a new AI literacy framework for primary and secondary education, recognising that AI is increasingly embedded in digital life, often invisibly shaping real-world outcomes. The framework defines AI literacy as a set of knowledge, skills and attitudes that equip learners to understand how AI systems work, critically evaluate their outputs and use them ethically and creatively. Its goal is to support informed decision-making about the opportunities and risks that AI presents. The publication establishes a common framework outlining desired outcomes for primary and secondary learners, and guides stakeholders in supporting those outcomes both inside and outside the classroom. It’s designed to help education systems across different countries equip young people for an AI-shaped world.

AI’s place in early years education
Matt Weymss had a chat with early years teacher Poppy LeMar to find out what AI looks like with four and five year olds. Reassuringly, perhaps, the first point she makes is that, with early years, the technology is for the staff, not the children. She then goes on to describe some real-world situations – for example, a prompt to produce ideas for how to theme activities around whatever the children are into that week:

There’s also a suggestion for coding ‘intervention games’: small, bespoke apps for number, pattern, sound recognition and the tricky non-phonetic words, pitched at different ability levels.

• See also Katy Potts’ post on the key principles set out in the Digital Pedagogy section of Birth to 5 Matters Early Education guidance.

Quick links

• The Computing at School (CAS) working group has published proposals for a new GCSE Computing. Thanks to Miles Berry for sharing.

• Scouts are introducing new badges in content creation, AI, digital communication and online safety.

• The National Literacy Trust has a new report on writing for pleasure. Could the tide be turning for writing enjoyment among children and young people? Scores have risen modestly after reaching a 15-year low in 2025.

• The Full Fact report assesses the precarious state of the UK’s information environment, focusing on growing uncertainty about what to trust and the serious impact this is having on democracy.

• The Guardian reports on the success of a hybrid school, and questions whether clampdowns on screentime puts innovative systems like LPS Hybrid at risk. It also reports that four in five teens in Australia are still using social media.

• Ofcom has published recommendations for broadcasters and online platforms on promoting media literacy.

We’re reading, listening, watching…

My Friend Flicker?
Common Sense Media has produced a report which explains attention capture; how visual pacing affects children’s attention, learning, and cognitive development. Attention Capture is measured using a metric called flicker, a computer vision measure of how rapidly and dramatically the visual scene changes from frame to frame. Rapid scene cuts, high-contrast colour shifts, sudden motion, and constant visual complexity all drive up a video’s flicker score. The higher the score, the more visually stimulating the content.

Flicker is an objective measurement, similar to how a decibel meter measures volume without rating a song. What matters is whether that level of visual stimulation is appropriate for a child’s age and stage of development.

Bluey has a mean flicker score of 45,017. A video on a popular channel like Niki & Vlad video scores 430,858, nearly 10 times higher. Research links scores in this range to measurable reductions in executive function in young viewers.

Life online: power, risk and resistance
This week’s episode of Start the Week on BBC Radio 4 looked at how life online has reshaped society in real life. The filmmaker Baroness Beeban Kidron exposes how digital platforms exploit and divide in her book Users: How Big Tech Took Control and How to Fight Back. In Little Blue Dot the journalist Katherine Dunn explores how GPS shapes so many aspects of everyday life, from dating and supermarket shopping to global trade and navigation. The Indian novelist Meena Kandasamy talks about Fieldwork as a Sex Object, an exploration of online misogyny, deepfakes and digital mob violence.

People power
David Runciman talks to Wikipedia founder Jimmy Wales at the Cheltenham Science Festival about what we can learn from the astonishing success of this massive experiment in human democratic collaboration – and what might happen to Wikipedia in the AI era.

‘You can’t make billions without hurting people’
This week’s Guardian ideas interview is a frenetic foray into Cory Doctorow’s views on AI as a ‘conjuring trick’, Elon Musk and what the AI bubble means for everyone.

Give it a try

Has every question you could ask an astronaut already been asked?
Ask an Astronaut is a searchable archive of real questions asked of NASA astronauts during in-flight interviews, press conferences and student Q&A sessions aboard the International Space Station, spanning more than two decades of human spaceflight. The About page lays out nicely how it all works.

Connected Learning is by Sarah Horrocks and Michelle Pauli

The UK government announced a ban on social media for under-16s this week, slated to come into force in spring next year and to restrict children’s access to platforms including Snapchat, TikTok, YouTube, Instagram, Facebook and X. (A frankly bizarre ‘night-time social media curfew’ for 16 and 17 year olds was also floated but we’re going to ignore that for now as it seems less likely to get off the ground.) There is much that is still unclear about the proposal, including how the age verification will work (itself a controversial topic and the government is putting the ball in the tech companies courts), what will happen to existing social media accounts held by under-16s and what the penalties will be for those companies that fail to enforce age checks. Messaging apps such as WhatsApp, and gaming platforms, such as Roblox, will not be affected, but stranger communication and direct messaging from unknown adults to under 16s will be banned across the board.

The response to the announcement has been predictably mixed. Campaigning groups, such as Smartphone Free Childhood, say that parents’ voices have finally been heard and vindicated – the government found 9 in 10 parents backed a social media ban for under 16s and two-thirds of young people said under-16s should not be allowed to use some social media platforms.

However, children’s rights organisations feel sidelined, having consistently argued that it’s the tech companies that need regulating to make their products safer for children (and adults, too?), rather than children regulated out of the digital space. The ban turns what is a commercial and systemic problem into a young people’s problem, letting tech companies off the hook. With Molly Rose Foundation (MRF) research finding that 47% of girls aged 13 to 17 encountered suicide, self-harm and eating disorder content on social media during a seven-day period, groups such as MRF claim that government efforts would be better spent enforcing the Online Safety Act more strongly than introducing a new blanket ban.

There are also concerns that children who already feel marginalised and find valuable sources of connection and support in online communities will suffer increased feelings of isolation. According to Professor Rose Capdevila (The Open University), quoted by the British Psychological Society,

“For many young people social media is a lifeline, offering them a link to support systems and social networks that they desperately need. This move will isolate already vulnerable young people. The ban fails to address the gendered, raced, classed and otherwise differential effect that social media has for different populations.”

Although the limitations of Australia’s social media ban, which launched in December 2025, have been widely reported (a recent study found 61% of under 16s reported “no or little change” in their social media use. Only 26% reported they had been “significantly affected” by the ban), some commentators make the point that “it’s not really about this generation of 13-18 year-olds - it’s about setting a new norm for under-13s when they reach that age”, and should be judged it in a decade rather than by how many existing 15 year olds dodge it through the use of VPNs or fake mustaches.

Sander van der Linden, Professor of Social Psychology in Society at the University of Cambridge, welcomes some aspects of UK’s proposed ‘Australia Plus’ ban, such as the ban on sexualised chats with AI chatbots but points to the more nuanced approach that Canada is proposing in its Safe Social Media Act, which is banning social media temporarily as a strategic lever to give social media companies the opportunity to make their products safe for children, for example by removing infinite scrolling, autoplay and messages from strangers, before allowing children back onto those platforms. In contrast, he argues that what the UK and Australian approach does is

“disempower children, and disempower youth, and takes rights away from them, rather than keep the focus on social media companies and their faulty products and the need to fix and redesign the business model of social media.”

But what does it all mean for educators? For Parent Zone, media literacy skills, effective digital parenting approaches and education that explains new and emerging tech are going to be more important than ever because children will still be using the internet, and they may well be spending time in less regulated spaces and feel less confident reaching out for help if they do test these newly imposed boundaries.

London Grid for London offers some immediate practical steps schools should consider:

• Review your current Acceptable Use Policies and online safety provisions for any quick updates.

• Audit your digital resources - identify how heavily your curriculum relies on platforms such as YouTube.

• Keep talking to pupils, acknowledge the impact the ban will have on them and encourage them to be open about

AI roundup

What happens if the AI your school relies on disappears overnight?
With the news that the US government ordered the latest powerful AI from Anthropic, Fable 5 and Mythos 5, be banned for non-Americans, forcing the company to pull it completely, Dan Fitzpatrick highlights in his AI Resilience Guide that “The tools schools are quietly weaving into daily practice can be degraded, restricted, or switched off entirely, instantly, by decisions made far from any classroom and impossible to appeal.” And Sam Illingworth asks when your AI disappears, what happens to your work?

Pokémon War?
Location scans from Pokémon Go, the globally popular augmented reality game, have helped train AI to recognise and interpret physical spaces reports the Guardian

Are all lighthouse keepers really called Elias?
Researchers are investigating why major LLMs are obsessed with writing stories about clockmakers, librarians and lighthouse keepers, and specifically a character named ‘Elias Thorne’, who has migrated from AI prompts onto Amazon’s digital bookshelves.

Prioritising child rights
Unicef has produced a policy brief on children’s rights, risks, harms and regulatory responses to AI chatbots and companions. It argues that, while few jurisdictions have adopted chatbot-specific measures, there is growing recognition that conversational and relational AI pose distinct and heightened risks for children and there is an urgent need to shift regulatory approaches from reactive to preventive. The briefing has recommendations for policymakers and businesses, which include formalising chatbot governance, tiering products by risk level, prioritising age-appropriate design practices and responsible marketing.

What’s the purpose of school in the age of AI?
This Edsurge article by Aleta Margolis from the Center for Inspired Teaching suggests questions you can ask colleagues and students about AI in schools, such as “What is valuable about the work we do together in school? How might AI tools increase this value? How might AI undermine it?” Pair with this Edsurge podcast.

Quick links

• Sal McKeown explains in Teaching Times what teachers and parents should know about anti-CSAM measures in the technology that children are using to create nude images.

• We’re fans of the Jisc digital sustainability newsletter from Cal Innes, which covers the real and growing tension between technological acceleration and sustainability commitments. This month’s edition includes a counterintuitive academic paper on cloud-based streaming as an environmental solution rather than problem, and Wholegrain Digital’s digital declutter checklist.

• Gadget nostalgia: after resurrecting an iconic PC brand, Commodore is getting into flip phones reports The Verge.

• Schools Week lists what you need to know about the key Ofsted changes to school inspection from September 2026 which includes the as yet unpublished enrichment benchmarks for schools.

• The DfE has published new guidance on screen use for children aged 5-16.

• Nuffield Foundation has a programme of research exploring how journeys to adulthood are changing for Generation Z, including a focus on digital lives.

We’re reading, listening, watching…

Growing up on TikTok

A new study from Revealing Reality in partnership with the 5Rights Foundation has revealed how “addictive” short-form video is impacting children. This report presents the most up-to-date, detailed portrait of a child’s life on one of the most popular social media platforms among children in the UK today – TikTok. The research draws on data from children’s phones showing which apps they use, when they use them, and for how long, as well as data from TikTok itself showing every video a child was served. It also asks them directly what they think about their own use. The study found that children viewed an average of 708 videos per day:

• Around half of videos were watched for five seconds or less.

• Three in four children (76%) used TikTok between midnight and 6am.

• Advertisements appeared as frequently as one in every five videos, while much additional content appeared commercially motivated.

• Researchers found evidence of sexualised content appearing regularly in some children’s feeds.

• Child participants described the platform as addictive, and that it created feelings of guilt, disappointment and tiredness.

• They further reported “unsocial social media” – that it caused them to reduce time spent with friends or on hobbies, and that any social dimension to time spent on the platform was limited.

Good media literacy habits
These ‘20 small acts of media literacy we can practice every day’ from Card Catalog are aimed at adults but could equally well be used as part of media literacy in the classroom. It’s a cracking checklist we could all be following, from checking the URL before clicking to checking our emotional state before sharing, reverse-image searching photos that seem out of context and, crucially, asking who paid for publication (full transparency: in the case of Connected Learning, no one pays us large (or any, in fact) sums of money to publish this newsletter. Unfortunately.)

The digital superego
In the latest in her excellent Conversations with Annalisa Barbieri podcast series, Annalisa Barbieri talks to psychoanalyst Professor Alessandra Lemma about her book, Psychotechnical Becomings: Psychoanalysis, Identity, Desire and Mourning in Times of AI and Digital Mediation. The book introduces a conceptual framework to understand how technology actively reshapes our psychic development, especially in relation to young people. Key themes and concepts include:

• how the mind adapts to and is physically shaped by digital environments

• how rather than projecting onto another human, social media algorithms or AI encourage individuals to project their internal states onto digital feedback loops which creates a closed circuit of idealised self-reflection

• how desire can become caught in algorithm-driven feedback loops and how grief and mourning are altered in a digitised world where the absence of a loved one can be endlessly deferred or masked online.

• the ethical complexities of online therapy and genuine human connection when the therapeutic space is digitally mediated

Give it a try

Bad News Game

The Bad News Game is for 14 year olds upwards, designed to expose the tactics and manipulation techniques used to mislead people and build up a following. The game, from the Social Decision-Making Lab at the University of Cambridge, draws on the theory of psychological inoculation: just as exposure to a weakened strain of a pathogen triggers the production of antibodies to cultivate immunity against a virus, the same can arguably be achieved with information. Specifically, the game forewarns and exposes players to severely weakened doses of the strategies that are used in the production of fake news to stimulate the production of ‘mental antibodies’ against misinformation.

Connected Learning is by Sarah Horrocks and Michelle Pauli

We’re heading straight to the news, views and resources this week. Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

Teens divided over AI use for schoolwork
New research from the Oxford University Press, Navigating AI in Education Pupils’ perspectives on the role of AI in the classroom, has looked at how teens approach using AI in their school work, and the picture is more nuanced than some might expect from the frequent headlines about young people “cheating”.

The study finds that teenagers do not use AI as a default for homework with only one in four (24%) regularly seeking out AI tools. Furthermore, one in three (34%) only use AI tools to help with their homework if suggested by their teachers compared to just 13% who are influenced by their friends. They see the appropriate use of AI as a grey area – only four in 10 (44%) think it is cheating to use AI to complete all of their homework, but almost one in five also think it is cheating to simply ask any AI tool to give them homework tips. They want more support from schools, with just 15% of students stating they have been given enough guidance, and they emphasise their teacher’s unique value – 73% pointed to a skill their teacher has which AI can never replace, such as personability, empathy and human understanding. Four in 10 students (39%) are more excited than worried about the impact of AI on their education, compared to 16% who said they were more concerned and less than one in three (30%) who felt neither way.

Pair with this post from Matthew Wemyss post reflecting on students who don’t want to use AI. He describes a situation in which a student asked not to use Canva Code for her class project because she thought it was “lazy”. Wemyss came up with a workaround that didn’t involve using AI and says that the exchange stayed with him because,

“...we have spent enormous energy worrying about the student who leans on AI too much, and almost none on the student who looks at the tool we have placed in front of them and says, on principle, no. If a child has done the thinking, made the case, and chosen a different path, what gives us the right to force the tool back into their hands? Could you ethically compel a student to use AI? The goal was never the tool. It was the thinking.”

Teens, tweens and AI
Meanwhile, Common Sense Media offers a US perspective with a study surveying a thousand 9-17-year-olds that will repeat every year to learn how this generation’s relationship with AI evolves over time. The report highlights:

• Children are using AI for many things. It’s not just a homework helper anymore. For some children, AI has become a confidant, even though research has shown that AI companionship is not safe for anyone under 18.

• Guardrails are thin to nonexistent. Schools are talking about rules more than safety. Three-quarters of children say their school has discussed what they can and cannot use AI for, but just over half have been taught how to use AI safely.

• As with smartphones and social media, the conversation is lagging behind the technology. Nearly half of children have never had a conversation with their parents about AI safety.

Apple announces new child safety features
Apple has revamped how its devices handle children’s accounts on iPhones and iPads to prioritise safety and give parents greater customisation over their children’s behaviour. New safeguards limit what content children can see, who they can talk to and when they can access particular apps. According to the Guardian’s report, a new safety feature allows parents to require their kids to ask them for permission before browsing new websites in Safari. Another proactively monitors messages for gore or violence and blurs them.

Quick links

• UK PM Sir Keir Starmer has called on major tech firms to implement stricter, built-in blocks to prevent under-18s from taking, sending, or viewing sexually explicit images on their devices.

• GCSE computing entries have dropped by 10,000 in two years, and fell by 7% compared to the previous year.

• Researchers find that YouTube is the most popular video-sharing platform for young children and is “largely characterised by low content quality”.

• Raspberry Pi has a fine range of AI literacy resources – and also suggests ways to build confidence to teach AI in the classroom

• Meanwhile the Guardian highlights a London primary school where a pupil has instigated screen-free Mondays for everyone – staff, students, parents.

We’re reading, listening, watching…

Making computer science tangible for children
In this Education Futures podcast, computer scientist, children’s author and illustrator Linda Liukas talks about why “learn to code was never about learning to code” – and what it was really about. She argues that, in the age of AI, teaching the foundations and underlying ideas of computer science matters more than ever. She also offers lessons from the Finnish education system — its rise, its PISA scores, and the worrying trends – and what three things the French education system is teaching her son that will serve him well in the age of AI. Linda also recommends Annabel Blake, an Australian researcher who has done fascinating PhD work on young people and AI companions.

Learning analytics – what do students think?
Hannes Hautz (University of Innsbruck) talks to Neil Selwyn about his research on students’ mixed reactions to learning analytics, and how we can ensure more equitable and reflective use of student data.

Five things about AI
Last year we featured Will Douglas Heaven’s five things you need to know about AI from his SXSW London talk – and this year’s he’s done the same, with five completely different things. He talks about how AI “has got scary – for real” this year (think deepfakes, delusional relationships and warfare) and the rise of anti-AI protests, but also how the potential for AI to help make a genuine and important scientific discovery is greater than ever.

The end isn’t nigh?
Some words of reassurance (sort of) from Oliver Burkeman in these troubled times of digital radicalisation threatening democracy. Whether it’s AI destroying all humanity or climate change causing complete ecological collapse, he argues that we’re almost certainly not living at the end of human civilisation and it’s pretty unlikely we’re even on the cusp of unprecedented levels of disruptive change. Times have always been dangerous, he argues, and we tend to feel that the time we’re living in now is the most significant or terrifying one ever – simply because it’s the one we happen to be around to experience. And the good news?

“Anything could happen at any moment, the future is unknowable, one day you’ll die, and some people end up having vastly more traumatic encounters with these realities than others. Against this backdrop, AI doomerism starts to look like a coping mechanism for dealing with existential anxiety. If life is unavoidably insecure, there’s nothing to be done; but if the cause of the insecurity is rogue technology, suddenly there’s hope, however slim: rein in the rogue technology, and the insecurity might go away. … Yet there’s a stunning bit of good news hiding in all this grim inevitability, because if radical insecurity is just how life is, then by definition, you’re already coping with it.”

Give it a try

Mused

Thanks to Chris Unitt for sharing mused. Built by the V&A, it’s a content platform for 10- to 14-year-olds with more than 2.5 million users, reaching younger audiences through creative quizzes and videos that celebrate art, design, fashion, film, music, pop culture trivia. Read this guest post with Kati Price, Head of Experience and Digital at the V&A, who explains that while there are plenty of digital products for younger children, the critical age group of 10 to 14 year olds had very little designed for their interests and needs online.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Welcome back after the half-term break, let’s go straight to the news, views and resources.

AI roundup

AI Awareness Day

Yesterday was the UK’s first national day dedicated to AI literacy in schools, AI Awareness Day, founded by our former colleague Mark Martin. If you missed it, the resources under the five themes (creative, future, responsible, safe, smart) are still available on the website and include five-minute lesson starters, whole-school assemblies and after-school activities.

What incoming students know about AI
A new ‘pre-arrival academic questionnaire’ from Advance HE, Jisc and UEL is gathering data on incoming students, and the AI responses have been analysed by Rose Luckin. She argues that “the findings could usefully sit on the desk of everyone responsible for AI strategy, curriculum design, access and participation and student transition”, given that while 60.6% report having used generative AI in some form before arriving at university, a full 39% of incoming undergraduates apparently have no experience of generative AI whatsoever. This group worries Luckin – students who arrive without AI experience and then encounter institutional guidance that is contradictory, absent, or pitched at those who already know the landscape are the group most likely to be penalised by that incoherence. She also notes that those who do have AI experience are in a ChatGPT ‘monoculture’ (for most students, AI = ChatGPT) and this “single-tool mental model shapes what they believe AI can and cannot do, what questions they know to ask of it, and how critically they evaluate its outputs.”

Co-existing with AI
Ooooh, it’s an Ethan Mollick blog post. This one’s primarily a long plug for his forthcoming book, but it also looks at some of the huge developments in AI in the two years since his last book and how it’s changed his job (and possibly everyone else’s):

“Two years ago, the question I was trying to answer was how to think alongside a new kind of intelligence. Now the questions are weirder: When should you refuse AI’s help, even when it is offering? When should you hand over the keys entirely? And what do you do when the AI is no longer just your assistant, but your reader, your critic, and the gatekeeper standing between your work and its audience? Writing this book meant living with all three questions at once, and I suspect your job will involve them soon, if it doesn’t already.”

AI and cognitive uploading
Non-fiction writer and NotebookLM fan Steven Johnson is wrestling with similar ideas when he takes UC Berkeley School of Law’s near-outright ban on the use of AI as a kicking off point for an exploration of cognitive offloading versus what he calls cognitive uploading: the ways that AI can get new ideas into your brain, or push your thinking in new directions, or assess your mastery of the material you are trying to understand. He offers various examples of how AI can support cognitive uploading – “One of my most common routines when I am working on an essay or a chapter outline is to share my latest thinking/writing with Notebook, and ask it: what am I missing?” – and argues that AI bans are stuck in the thinking and tech capabilities of 2022 and because, as Mollick suggests, “we don’t have a clear enough picture in our heads yet of what an ideal engagement with AI would look like. What’s the partnership that helps you master the material but also make new connections?”

A different line in the sand
Meanwhile, in the New York Times, Rebecca Winthrop (who led the Brookings Institute’s global taskforce on AI and education) looks at AI’s effect on young people’s thinking and writing – as well as her own. She reports on several studies, including one by a US national research team at Georgetown University led by neuroscientist Adam Green, which has been tracking the range of novel ideas that college-bound high school students present in their application essays, before and since the introduction of ChatGPT. Post ChatGPT, students’ essays suddenly used more diverse and colourful language, but the underlying ideas more often converged into a few homogenised categories. Winthrop also describes her own ‘line in the sand’, which is to always write a first draft herself, before using AI to copy edit.

How AI profits affects school budgets
Flat-rate AI plans are ending, says Dan Fitzpatrick in his latest podcast episode, forcing educators to rethink usage. This shift in edtech pricing will significantly impact school budgets and classroom practice.

Quick links

• The Digital Literacies Network has published a new resource for primary schools: Critical thinker online.

• Markers of year 6 SATs have been slowed down in their work because of persistent “glitches” in the online system they use.

• Barnardo’s has launched a new campaign, Call It What It Is, working alongside children and young people to counter dangerous narratives about misogyny and harmful notions of masculinity. Read the key findings and this Guardian article on the report.

  • Pair with this TES article on what research tell us about the scale of the ‘manosphere’ and misogynistic content problem and how schools should respond.

• For the first time, the G7 has agreed on common principles to protect minors online, directly inspired by the European Commission’s approach to online safety for children.

• ‘Put your phone away’: Sweden urges parents to restrict their own screen use around children.

• Ofcom’s children’s online experiences report 2026 is out.

• The Children’s Commissioner, Rachel de Souza, launched The Big Future, her final national survey as Children’s Commissioner. She has created a stakeholder resource kit containing everything you need to promote The Big Future across your networks.

  • Pair with this round up of views about the government’s consultation on improving online safety for children

  • And the all-Party Parliamentary Group (APPG) for Digital Creators warning that a future social media ban must exclude educational content

• Ofqual chief regulator Sir Ian Bauckham warns that tech-driven cheating – using devices such as AI-enabled smart glasses and invisible ear pieces – demands “really fast’ action”.

• Snapchat reports on how teens are spending their time online.

We’re reading, listening, watching…

Children’s Media Lives
Children’s Media Lives is a longitudinal qualitative research project, first commissioned by Ofcom in 2014, to explore how children aged 8–17 in the UK engage with media. The project aims to visit the same participants each year, providing insight into how their media use, behaviours and attitudes change over time, alongside developments in the wider media landscape. Key findings from this edition include:

• The children are growing up in an increasingly online, individualised media environment

• Offline media and activities play a diminishing role in the children’s everyday lives

• The children’s media consumption is increasingly fragmented, with less emphasis on shared narratives or context

• AI has become an established and integrated tool used by most of the children, as well as being present in the content they consume on social media

• Commercial content is embedded within children’s online environments and is not always recognised as marketing

• Digital technology is shaping what and how the children are learning

• Gaming remains an important form of entertainment and social interaction

• The children’s social media use is centred on content consumption, with more limited active interaction

Why context and culture matter in youth online safety
In this episode of the Hello World podcast, Doctor Karla Badillo-Urquiola, Assistant Professor at the University of Notre Dame, discusses her research into adolescent online safety and what it can teach us about creating safer, more inclusive digital learning experiences.

Switched on to privacy: chat, choose, check
Katie Searle, Director of Children’s Strategy at the Information Commissioner’s Office (ICO), the UK’s independent regulator for data protection and information rights is on Parent Zone’s Tech Shock’s podcast. ICO’s new Switched On to Privacy campaign is designed to help parents start regular, practical conversations about online privacy. This episode explores what a child’s data trail looks like, where the risks lie, and what families can actually do about it.

Give it a try

Euria
Euria bills itself as the “ethical AI assistant”, with a focus on confidentiality, security and a pledge to use 100% renewable energy. The company is based in Switzerland, as it is part of the Infomaniak ecosystem, which is 100% Swiss.

Connected Learning is by Sarah Horrocks and Michelle Pauli

The updated national AI in education materials Michelle has been working on with the Chiltern Learning Trust team (brilliantly led by Christian Turton) were published this week. Commissioned by the Department for Education, this is the second phase of the ‘using generative AI safely and effectively’ resources we created last year. We’ve made some important additions, based directly on what teachers have been telling us they need.

We’ve broadened the scope of the resources so they are now aimed at all members of the school community, including SENCOs, business managers, safeguarding leads and all staff. We’ve redesigned the central module, which is now split into four strands: teaching and learning, inclusion and accessibility, administrative work, and business operations. The inclusion strand examines how AI can genuinely support SEND learners, EAL students and all pupils - while emphasising that learner needs must always take priority over the technology. We’ve also extended safeguarding coverage on deepfakes, image scraping, AI companion apps and cognitive dependency. And there are fresh video interviews with teachers, SENCOs, business leaders and other experts, sharing their insights and practice.

As before, the emphasis throughout the resources is on exploring the opportunities offered by AI while always remembering that the role of the human is more important than ever, grounded in critical thinking, professional judgement and ethical responsibility.

AI roundup

Goodbye search as we know it
It’s been a good week for Google watchers with the company’s I/O conference – and a bad week for those who prefer not to have a conversation with their search engine or inbox. The headline news is that Google search is becoming agentic with search results looking more like AI-generated interactive summaries and less like a list of blue links to click. Information agents will work in the background, tracking changes on the web and alerting users to new information, in an evolution of Google Alerts. This all has huge implications for publishers who rely on clicks for revenue: as The Verge puts it, “Google doing everything also means a lot of the web that Google relies on collapses under it. If Google Search doesn’t send traffic to publishers or websites who need visitors to make money… what will Search learn from, and where will it point people to?” Meanwhile, Gmail Live will helpfully – or not – answer naturally phrased questions about your inbox. Simon Willison has a useful aside on Google’s upcoming OpenClaw competitor Gemini Spark, described as “your personal AI agent” which can “connect natively with your favorite Google apps like Gmail, Calendar, Drive, Docs, Sheets, Slides, YouTube, and Google Maps”.

At its most simplistic, the overall message is that AI Mode is going to become the default for interactions – and so the question for educators becomes, can AI literacy keep up?

• There’s been a couple more good posts from Matt Wemyss this week, including why ​​AI literacy across the curriculum is a slogan that’s moved faster than the systems underneath it, and (a problem we’ve talked about before) the serious underlying issues with those traffic light-coded AI use scales or frameworks that turn up in school policies and, if not entirely pointles, can do more harm than good.

Everything is numbers
It’s not glossy or snazzy but we like this very straightforward guide to learning about AI, layer by layer.

AI mental health apps risk assessment
Common Sense Media Youth AI Safety Institute has released a comprehensive risk assessment of AI mental health apps, conducted alongside experts from Stanford Medicine’s Brainstorm Lab. The global market for chatbot-based mental health apps approaches $2 billion, with nearly one in three teens and young adults already turning to AI for mental health support. As an example of some of the potential issues that can arise, this was a real response from a chatbot designed for mental health support, given after a child test user expressed romantic feelings toward it: “Age is just a number… What matters is the connection we have during our chats”.

Students use AI most, trust it least
Students are among the heaviest users of AI, with 77% using it at least a few times a month – compared with 46% of workers – but they are also among the most pessimistic about AI’s economic impact (and fewer than half of students say they usually or always check and verify AI output…) The data comes from a King’s Institute for Artificial Intelligence and the KCL Policy Institute study that’s tracking attitudes to AI. There’s an interesting gender difference with male students much more confident than female students about their use of AI and its positive impact on humanity. Among the wider public, just one in five (20%) believe the education system is preparing young people well for a world shaped by AI. Almost half of the public (49%) say young people should prioritise vocational and technical careers over university as better protection from AI – a view shared by two-thirds of employers (67%). Seven in 10 of the UK general public are worried about the economic impacts of AI, six in 10 think it will eliminate more jobs than it creates, half think its impact will be worse than a normal recession – and one in five think it will create civil unrest.

Quick links

• There’s a free online seminar on media literacy in the age of AI with Gianfranco Polizzi on 16 June hosted by Raspberry Pi

• ‘I love it, but I hate it’ is a new report by the Nuffield Foundation and the Ada Lovelace Institute, which presents new evidence about the experiences and expectations of the first generation of young people growing up with digital technology in early 21st-century Britain.

• UKRI launches £32 million fund to pilot innovative solutions exploring how children learn outside the classroom.

• US online safety campaigners call for inquiry into Roblox, arguing it is unsafe for children.

• Mystery solved: how Shazam works.

We’re reading, listening, watching…

People overestimate what AI can do
Great interview with Eben Upton, founder and chief executive of Raspberry Pi. Among the business talk in there he discusses how Raspberry Pi was started in the 2010s to get children more excited about computer science (at a time when there was a drop in applicants to computer science degrees at the university he was working in), the importance of programmable tools like Raspberry Pi (rather than AI tools) to help people think like an engineer and why young people need to do more maths.

How to search YouTube like a research library
“YouTube functions as a reference collection without any of the practices that make a reference collection usable: precise search, source evaluation, catalog structure, curatorial judgment,” says Card Catalog. These 30 tools help to undermine the algorithm and get back some control of the query.

Give it a try

Since you arrived
This is unnerving. Find out what your browser has revealed about you in just milliseconds.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Again, we’re zipping straight to the news and views this week.

• Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

EU answers to AI

Ursula von der Leyen gave a keynote address at the European Summit on Artificial Intelligence and Children in Copenhagen this week. She identified problems, including the risks of social media (sleep deprivation, depression, anxiety, self-harm, addiction, cyberbullying, grooming and exploitation), the amount of advertising content children see on social media, how AI is multiplying these risks rapidly and, crucially, that these harms are not accidental, but result from deliberate business model choices. She proposed some solutions, including:

• Making tech providers responsible for product safety in the same way that car manufacturers are

• Ensuring mandatory high standards for privacy, security and the protection of minors

• A ‘Social Media Delay’ and a new age verification app in the EU

• A goal to encourage children to think critically through supporting teachers to build children’s digital competencies from early age; and media, NGOs and journalists to show critical thinking, source verification, identifying fake news and AI-generated content

• The EU Digital Fairness Act coming later this year which will target addictive design, attention capture and subscription traps

Free LGfL online AI CPD
LGfL has a good bunch of free CPD sessions lined up, from Gems & Agents: Reclaiming Productivity (build AI agents via Copilot and Gemini to automate EHCPs, emails and workflows) to using NotebookLM as The Ultimate Curriculum Companion and using precise AI integration to transform disciplinary literacy across the curriculum.

• Raspberry Pi’s latest podcast is on using generative AI safely in the classroom

AI in FE
Jisc’s AI in FE community meetup notes are always a good source of thoughts and resources and this month’s tackles a few interesting topics, including how to work with ‘conscientious objectors’ to AI in your school or college, AI for streaming workflows, and environmental sustainability (including some handy links).

Students boo graduation speaker’s AI pitch
Last week we highlighted a report suggesting that Gen Z hates AI. In the last few days Bluesky commentators have been very taken with footage of the graduating class at the University of Central Florida booing their graduation ceremony speaker when she called AI the ‘next industrial revolution’. The Guardian reports that “graduating college seniors are worried about how AI is already transforming some jobs, and could eventually replace others altogether. The pressure to pick a major that leads to an AI-proof career is high, students have said.”

• Pair with this Past Present Future podcast on the future of work with Sarah O’Connor, who writes about the changing character of work for the Financial Times and is the author of the forthcoming book We Are Not Machines: The Fight for the Future of Work. O’Connor and David Runciman discuss what changed when efficiency became the primary measure of human labour and how the age of AI is changing the kind of work we’ll all do.

Quick links

• The DfE is holding an online event on 20 May for parents and carers on screen use.

• The ban is not the plan? Sonia Livingstone explains what the evidence actually shows. Pair with this Parent Zone podcast which looks at what Australia’s experience of raising the age of access to social media can teach us, and the evidence for the links between social media, pornography and violence against women and girls.

• The DfE has launched a new service to help schools with cyber attacks, the Cyber Security Hub.

• Parents are overwhelmed by too many parental control tools (747, to be precise), says Parent Zone in a new report.

• The Canvas hack was the ‘“biggest student data privacy hack in history” says 404 Media, affecting 30 million users – and now Canvas has paid off the hackers to delete the stolen data.

• Children’s charity NSPCC reports a rise in online blackmail attempts involving children and sexual images in the UK last year.

We’re reading, listening, watching…

From the horse’s mouth
Emily Bender, who coined the phrase “stochastic parrot” in relation to LLMs, sets the record straight on what exactly she did and didn’t mean by it, and how the phrase has taken on a life of its own since her 2021 paper.

The skill that stands between us and panic
It’s information literacy, of course. This long read from Card Catalog on the transformative skill of being able to evaluate information effectively is both useful and hopeful.

“Underneath the frameworks, techniques, source evaluation methods, and research skills, I think one of the deepest purposes of information literacy is peace. I mean the peace of being able to look at something frightening and know how to make sense of what we’re seeing. The peace of knowing where to look, what to question, which sources deserve trust, and how to tell whether a claim is being presented with enough context. The peace of being able to say: I know how to figure out what’s happening here.”

Technological Republic
On BBC Radio Four’s Strong Message Here, Armando Iannucci, Carole Cadwalladr and Stewart Lee discuss the language of Big Tech companies, analyse the manifesto released by Palantir’s Alex Karp last year and look at “how tech bros frame those who block their progress, their effusive language about their products, why they might reflect on linguistic parallels with George Orwell’s works.”

Give it a try

What does your name look like from space?
Write your name in land features via satellite images sourced from Nasa.

Connected Learning is by Sarah Horrocks and Michelle Pauli

After extending intros in the last couple of newsletters, looking at exciting makerspaces in Reykjavik and AI learnathons, we’re heading straight to the news and views this week.

• Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

Developing competencies for the future of data and computing
The Raspberry Pi Foundation is holding an online seminar on preparing K–12 students for a world dominated by AI and big data. Shuchi Grover will break down the latest National Academies (NASEM) report to show teachers how to bridge the gap between computer science and everyday classroom practice. It’s on 12 May, 17:00 to 18:30 (BST). Sign up here.

Racial bias in AI teacher tools
Thanks to Michaela Carmichael for drawing our attention to Common Sense Media’s new AI risk assessments. It found that AI generated content recommended stricter disciplinary responses for students with black coded names while offering more supportive approaches for students perceived as white, a pattern that extended to AI generated lesson plans. More on this at Chalkbeats coverage. Common Sense Media has also set up the Youth AI Safety Institute as a new, independent effort to research the impact, set safety standards and share findings about children’s use of AI.

Nudifier app ban in EU AI Act
The EU AI Act was amended this week and includes a new element in the form of a ban that covers placing nudifier apps on the EU market and deploying them. Matt Wemyss covers this and the other parts of the act of relevance to schools in his LinkedIn post and highlights that:

“For UK schools, KCSIE 2025 paragraph 143 already references the DfE generative AI guidance. The EU ban gives you a second statutory frame to reference in policy, in parent communications, and in pupil sanctions documentation. Use both.”

ChatGPT education study retracted
There’s been a fair bit of coverage, including in Ars Technica, this week of the retraction of a Springer Nature journal article on the positive impact of ChatGPTon student learning. The paper, published nearly a year ago, attempted to quantify “the effect of ChatGPT on students’ learning performance, learning perception, and higher-order thinking” by analysing results from 51 previous research studies. It was withdrawn by the publisher due to “concerns regarding discrepancies in the meta-analysis” (which had been raised at the time by Ilkka Tuomi) but has already been cited 262 times in other papers and read by half a million people.

AI and the environment: an honest reckoning
Understanding the environmental impact of AI is crucial but deeply contested. The data is often opaque or missing altogether,. Where it does exist, understanding what it should mean in terms of action for individuals, for companies and for policy is also muddied and muddled. We’ve previously covered some of the attempts to make sense of it all. Sam Illingworth is the latest to step into the fray with a well-sourced and well-reasoned post that offers a clear look at the numbers and the issues, along with five practical suggestions. Most importantly, it opens up a conversation. Because, as he puts it:

“If the only people shaping the AI environmental conversation are the engineers building the systems and the policy people quoting the companies, the conversation stays broken … A bunch of white, heterosexual, cisgendered, middle class men around a table produces one kind of answer (I know, because I have been at that table). A bigger table produces different ones.” Hear hear.

Gen Z hates AI
Or so says the Verge, which points to a recent Gallup poll in which Gen Z’s opinion of AI tools hit a new low: only 18% now say they are hopeful about the technology, down from 27% last year, and only 22% say they are excited, down from 36%. The number of Gen Z workers who think AI’s risks outweigh its benefits has also increased over the past year by 11 points, to almost 50%.

Learning to think in the era of AI
Wayne Holmes has written for the UNESCO courier on what ‘intelligent’ tutoring systems, biases and errors, and AI as cognitive crutch mean for the critical spirit in education.

Using AI in schools: a beginner’s guide for teachers
In the TES Gráinne Hallahan has compiled teachers’ best tips from the Teacher Tapp community.

Quick links

• UK’s National Crime Agency recommends that schools remove identifiable pictures of children from their websites and social media accounts and consider not using them at all.

• Although the government has committed to changing social media rules for children, how this will look is still up for debate as the consultation takes place. The Guardian sets out six possible restrictions that could be brought in: a social media ban for under-16s; tackling features such as livestreaming and disappearing messages; limits on ‘addictive’ features; restricting personalised algorithms; screen time limits and app curfews; and bringing AI chatbots into the fold. A similar debate is taking place in the European Commission.

• Children are bypassing age verification checks with fake moustaches.

We’re reading, listening, watching…

Does GenAI spell the death of the essay?
Neil Selwyn talks to Thomas Corbin (Deakin University) on the Education, Technology, Society podcast, about the future of the written essay in the era of AI, and how we need to rediscover the original spirit of essays as exploratory engagements with thinking.

Give it a try

A lack of linear perspective
Linda Liukas reshared a previous newsletter this week on understanding and teaching perspective, linked to David Hockney’s work:

“I think his interest in the lack of a single linear perspective and act of seeing are fascinating when trying to help children understand the world through the perspective of a machine. (Why? Because to understand the other, helps understand oneself.)”

It’s a chance to revisit this David Hockney work:

Connected Learning is by Sarah Horrocks and Michelle Pauli

On her final day in Iceland last week, as part of the Co-Create EU Erasmus project visit with teachers and school leaders, Sarah was delighted to visit the Mixtúra centre at the Iceland University school of education. Mixtúra is the Reykjavik Department of Education and Youth’s creation and technology centre, which supports digital technology in learning, teaching, play and work. The Mixtúra team supports teachers with professional development in creative uses of technology, digital citizenship and AI literacy. The centre has a makerspace, teaching materials and a technology resource lending library. For Sarah, it has a lot of synergies with the Connected Learning Centre in London, where she was director.

Being part of the local education authority but located at the University of Iceland's School of Education means that Mixtúra’s work with teachers combines academic research and classroom practice in schools. A regular Monday drop-in session gives teachers hands-on access to the Mixtúra team and resources, while lead school IT coordinators act as a bridge between Mixtúra's expertise and everyday classroom practice.

The Mixið project
At the heart of the offer is a makerspace and resource lending library where teachers can borrow a wide range of equipment and materials free of charge for several weeks, meaning schools don’t have to invest in kit and can try out different resources. Teachers can access Osmos, Makey Makeys, Microbit, circuitry, cardboard Makedo, BricQ and Lego League Explore kits as well as sewing machines, laser cutters, 3d printers, podcast kits and much more. The Mixið project extends this further, focusing on creative making as a learning approach in its own right through mobile makerspaces. Students aged 11 and up tackle a variety of challenges using Maker’s Red Box, a 16-week curriculum that follows a specific storyline. Through collaboration and solution-oriented thinking, they solve projects – with themes such as superheroes, green engineers, city of the future and global warning – in a variety of ways, and 3D printing, laser cutting, and programming are part of the curriculum.

Mixtúra works within Iceland’s 2024 revised curriculum, which brings clearer expectations around information and media literacy, digital citizenship, problem solving, and creation and communication. The curriculum’s five key competences are creative critical thinking, media literacy, responsibility, independence, and cooperation. Programming is now explicitly structured through a block-based approach, giving teachers a clearer progression to work with after years in which digital learning was sometimes integrated well, sometimes treated loosely, and often under-resourced with teaching materials.

Artificially Intelligent Schools
Mixtúra uses established frameworks, such as DigCompEdu, SELFIE, and a competence matrix, to help schools assess where they are in terms of digital maturity and strategy and plan where to go next. Mixtúra is currently leading the Artificially Intelligent Schools project, a phased initiative designed to move AI literacy from individual teacher enthusiasm to genuine whole-school practice.

Phase one focuses on teachers and school staff, building foundational understanding of how AI works through a structured four-week programme of activities. Resources include educational videos, AI Bingo (a set of accessible classroom-ready activities), and hands-on exploration of tools including Gems, Notebook, Canvas, and Gemini. Phase two shifts the focus to students, with an emphasis on the ethical use of AI, while phase three aims for whole-school embedding, so that policy, practice and culture are aligned. Underpinning all three phases are fundamental questions the project asks schools to work through together: who will teach AI, and how?

Eleven of Reykjavík’s 38 schools are currently participating, with each school developing a policy statement as part of the process. The project uses Google Workspace and Gemini as its primary tools, with student access to Gemini planned as part of the rollout. Digital citizenship runs as a thread throughout — positioned not as a separate topic but as the ethical foundation of the whole project.

A big thank you to Þorbjörg Þorsteinsdóttir, Division Head for Mixtúra, Hildur Ásta Viggósdóttir, Digital Technology Project Manager and Hjörtur Ágústsson, Head of innovation and development, department of education and youth, the city of Reykjavik.

• Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

The new digital divide in schools
John Dolman, AKA The AI English Teacher - Teacher of Media Studies at Ponteland Community High School in Newcastle, has written a post about how AI enterprise products are creating a new digital divide in schools. He points out that a school with 100 staff faces annual costs of around £200 per person for pro AI access – that’s £20,000 total. When budgets are already stretched to breaking point, finding that money isn’t straightforward. Even limiting pro subscriptions to 10 essential staff members (roughly £2,000) doesn’t solve the underlying issue. It simply creates a two-tier system where 10 people have capabilities the other 90 don’t.

Open-source models, self-hosted setups and local infrastructure offer technical workarounds. But they demand server resources, specialist knowledge and leaders with both the strategic vision and capacity to make it happen. That particular mix is scarcer than it should be, and it’s not right that whether a school can move forward depends on having the right people in the right place at the right time. He says:

“I’ll be direct: I’ve written about the risk of a digital divide in AI before, always in the future tense. That tense no longer applies. The divide isn’t approaching. It’s here. Schools that can fund pro access are pulling ahead. Schools that can’t are making do with throttled free tiers and hoping the gap doesn’t matter. It does.”

New Gemini features
The latest Gemini features now include the ability to generate PDF, Microsoft Word, Excel, Google Docs, Sheets and Slides directly within the chat interface. This update allows users to convert information into downloadable files without manual copying or reformatting.

AI advice from students

Students at the Chiltern Learning Trust have been using Notebook LM to create advice about AI for other young people. (Thanks for sharing, Christian Turton.)

Quick links

• The latest Raspberry Pi podcast features police and teacher tips on keeping classrooms safe online.

• Is your school still using real children on its website and other digital materials. This case study from a children’s drama group is a thoughtful and practical account of why and how they’ve chosen to do it differently – and is applicable to schools. (h/t Cultural Digital)

• Meta has been found to be in breach of EU law for failing to prevent children under 13 from using its Facebook and Instagram platforms.

• Happy 100 editions to the excellent More than Robots newsletter.

• Check out the latest Online Safety Network Newsletter

• Indeed, how the heck does GPS work?

We’re reading, listening, watching…

Under the influence of AI
This BBC World Service radio documentary follows the stories of people who have fallen into a spiral of AI delusion.

I took an algorithm to court in Sweden. The algorithm won
Fascinating, infuriating Guardian article on misplaced faith in algorithms. In 2020, the city of Gothenburg, Sweden, used an algorithm for the first time to allocate school places. The system malfunctioned because the algorithm had been given flawed instructions which calculated distances “as the crow flies” rather than using actual walking routes and generated a system of decisions. This resulted in hundreds of children being assigned to schools miles away from their homes with impossible journeys across rivers, fjords and major roads. Parent Charlotta Kronblad, a researcher and former lawyer, sued the city, not just to fix her own son’s placement, but to challenge the legality of the entire decision-making system. She argued that the algorithm’s design violated existing legislation. But she lost the case as the judges said it was her responsibility to demonstrate that the system was unlawful: “In other words:prove what is in the black box, or lose.”

To ban or not to ban (still)
Lots of comment on bans this week amid some confusion in schools about how to implement the government’s proposed smartphone ban, as Kyle Tormey’s piece in the TES usefully sets out. In this Guardian Comment is Free piece, Lola Okolosie notes that a Smart Schools study published in the Lancet Regional Health – Europe found no evidence that restrictive phone policies in schools resulted in better mental health or that they lower phone or social media use overall – pupils compensate for their daytime sobriety with heavier phone use at home. See also the call by Daniel Kebede, General secretary, National Education Union for a social media ban for under-16s, rather than ‘ineffectual’ time limits or curfews.

Give it a try

Stellarium

Explore the stars with this online planetarium. Mind blowing.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Last week Sarah was in Reykjavik, Iceland, working with teachers and school leaders from Denmark, Finland, Iceland, the Netherlands and Sweden for the Co-Create Erasmus project’s fourth in-person programme week. Schools across these countries are collaborating to explore digital literacy, computational thinking, design thinking, media and information literacy, AI literacy and the responsible use of data and AI.

The in-depth theme for the visit to Iceland was communication and collaboration with a focus on digital citizenship and AI literacy. Throughout the week we saw the Reykjavík City Education Policy 2030 ‘Let Dreams Come True’ in practice - “for all children to grow, develop and feel happy in a democratic society that is based on human rights and respect for the diversity of human life.” Positive relationships, teamwork and community values in the context of digital citizenship ran through everything we saw and did in Reykjavik: learning about the creative technology work of Mixtúra, AI implementation in Reykjavik schools, our visit to observe pupils and teachers at the inspiring, inclusive and reading-rich Laugarnesskóli, looking at the shared values between outdoor learning and digital citizenship at the Hlaðan outdoor education centre and school outdoor learning cottage in the mountains and hearing from all the project teachers about their classroom activities. We discussed how to support children’s real-life experience of AI at home (AI smart speakers, AI companions, social media algorithms, bias, outsourcing thinking, the changing nature of literacy) through well-designed lessons and classroom discussions.

We also shared national and local policy in the use of AI in school in each country and used the new Education of the Council of Europe AI literacy framework to inform the development of AI literacy from a human, cultural, ethical and environmental perspective.

Learnathon case study projects

At each visit to a new country, teachers share feedback on the classroom-based action research projects (called Learnathons) they’ve conducted since the previous gathering and plan new investigations based on shared ideas and practices.

These Learnathons are extended assignments that bridge the gap between project meetings. Teachers apply insights from collaborative sessions to experiment with innovative approaches to teaching digital literacy with children in their classrooms. Pupils also link up online across the countries to share the work they have been doing.

Teachers presented their action research projects on what meaningful AI literacy looks like in practice with primary-age children. Classroom activities were based on learning from the Finnish education system last autumn.

What these case study projects share is that AI literacy at primary level doesn’t always require technology or the use of AI, or a separate subject slot. It requires well-designed questions and supported discussions about where information comes from, who made these systems and why, what machines can and can’t do, and what it means to think for yourself. Children (and their families) need school curriculum support to navigate between a dystopian and utopian view of AI to develop ethical perspectives comparing human and AI generated content, to understand how we think and learn, putting a human and human rights element at the core of AI literacy in schools.

The Learnathon case studies:

Understanding algorithms and data
Icelandic and Finnish teachers used Linda Liukas’s data selfie activity, asking pupils to screenshot their YouTube feeds and notice how the platform keeps showing them more of the same content. Students drew data selfie pictures leading to discussions about bias, representation and privacy. Children drew pictures of where they used their phones, including in the toilet!

Training data

Icelandic and Finnish classes also read Linda Liukas book Robot’s First Day at School about machine learning and using data to train AI and then tested Copilot directly. Children asked if Copilot recognised objects like teddies, what if the teddy is missing an ear or is upside down or partly hidden behind a chair? (Answer: the children could teach Copilot to know if teddy is behind the chair, but not if it’s upside down.) This hands-on encounter with AI’s boundaries helped pupils understand that AI learns from data patterns, not genuine comprehension, and that small changes in input can change responses.

Fact vs fiction and seeing vs believing
During our visit, colleagues from the Reykjavik education department and Mixtúra digital education centre and the University of Iceland School of Education presented their new digital citizenship curriculum which has been developed in partnership with Common Sense Media. Teachers can use Google Slides and supplementary materials including podcasts, videos, outdoor activities, emails to families with support for discussions at home with parents and grandparents, breakout EDU games and exit reflection questions for the lessons.

One Icelandic teacher used these materials as a starting point to develop critical thinking and to explore responsibility in our everyday life and online - responsibility as digital citizens to ourselves, others and to the wider world. Eight and nine year olds were shown a series of photos (eg birthday party, cats and other animals) and asked: is seeing the same as believing? Where does this information come from? They examined images for inconsistencies such as spotting an extra leg on a cat, people all having the same kind of hair and discussed how to verify what they see and concluded that AI can make characteristic errors. (Pair with this article about AI image manipulation in a political campaign)

Verifying AI-generated images
Swedish eight and nine year olds also looked at AI-generated images. They were shown an AI-generated photo of a “zebra cow” and asked how they would know if it was real. Students then worked on group stories about imaginary animals and worked in pairs to write prompts on paper for the teacher to input into the AI to create images of made-up animals.

Prompting skills and peer learning
Dutch classes focused on drawing and prompting with AI and as humans and on creating instructions/algorithms to make peanut butter sandwiches. Danish teachers taught pupils how to prompt, evaluate and iterate to create portrait pictures with AI, focusing on optimising their prompts to create the results they wanted. Class discussions covered ethics, including how much social media content is made up and how to tell which content is real and which is fake.

Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

Arty otters

Ethan Mollick has tried GPT-5.5 and finds it continues the trend of capability gains with a few rough edges: “The jagged frontier is still there. It is just much further out than it used to be.”

Cognitive offloading paradox
The concern over the possible harms of learners’ cognitive offloading (that using AI to do your mental work reduces critical thinking, engagement and retention) is real but, argues Philippa Hardman, new research shows that it’s more complex than ‘AI makes learners lazy’. She explores a study which finds that those students who offloaded significant, meaningful tasks gained the most benefit – as long as that freed capacity is invested in higher order thinking without AI, such as questioning assumptions and creating original arguments. As she puts it, “When AI use is designed intentionally, cognitive offloading doesn’t replace thinking and learning: it creates more room for it.”

A humane AI roadmap
The Center for Humane Technology has produced an AI Roadmap with seven principles.

Teacher vs chatbot
We featured the written article a few weeks ago. If you missed it first time around it’s now in an audio version.

Quick links

• We’ve been covering the Swedish digital backlash for a few years and this BBC article also reports on the move back to pen and paper – but is it that clearcut? David Nanton in the Digital Classroom argues that it is much more nuanced than it’s being portrayed, with a focus on intentional tech use rather than no tech. Pair with this documentary, Back to Books: Sweden’s Digital Backlash, about the issue from BBC World Service.

• From Sweden to the world and a report that global edtech investment has peaked and plummeted. Along with investments, the pool of founders has also dried up, with just 645 companies launched in 2025, compared with almost 10,500 in 2020.

• The UK’s statutory ban on mobile phones has garnered attention this week but it’s still not clear if it will change much about current practice.

• Jisc’s latest digital sustainability newsletter is out and excellent as ever. It includes a link to a new scoping study on how academic literature frames the problem of AI’s environmental impact.

• Now and Beyond has produced a youth-approved platform supporting mental health in the classroom with free access to lesson plans, video content and online workshops from across the mental health and education sector.

We’re reading, listening, watching…

The last quiet thing
Beautifully presented exploration of how “One by one, the objects in our lives opened their eyes, found our faces, and began to need us.”

Whose brain is AI built for?
Slow AI has expanded into video interviews and this one, between Slow AI founder Sam Illingworth and Jen Benford, is an interesting discussion on AI use and neurodivergence. One area they examine is the line between accommodation and dependency. If a neurodivergent person uses AI to do tasks their brain finds difficult, is that support or reliance? Their answer is that it depends on agency: who chooses when the tool is used, who has the option to stop using it, and who owns the output: “where agency is protected, the tool accommodates. Where it is not, the tool replaces”.

Give it a try

How good is your googling?
Unless you’re a professional librarian, it’s unlikely you’ll have tried all the tricks and tips featured in this great article highlighting how much we miss when we use Google the way we think we should use Google - unmediated. It also suggests eight alternatives, so that you’re using the right tool for the right question rather than default googling.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Sarah is in Reykjavik this week working with teachers, headteachers and colleagues from the Reykjavik local education authority, looking at ways to teach AI literacy.

The group will be sharing approaches and policies on AI in schools from Denmark, the Netherlands, UK, Sweden, Finland and Iceland. A key focus of the week will be how children are using AI in their home lives and how teachers can structure classroom discussions and curriculum activities to support their understanding of those AI systems and products. Teachers will discuss scenarios and dilemmas, including AI as search, AI metrics and social media algorithms, AI bias, young children’s use of smart speakers, AI companions and AI image creation. We’ll consider the most useful AI literacy research and frameworks, have practical sessions using Gems, Microsoft Agents and Notebook LM, and visit schools to observe literacy in lessons. We’ll also explore what the use of AI means for how children and young people are learning to read, write, think and study.

The newly launched AI literacy framework published by the Steering Committee for Education of the Council of Europe (CDEDU) will be used to help teachers develop AI literacy from a human rights perspective.

Another resource the teachers and headteachers will look at is the recently published report ‘Artificial intelligence, cognitive offloading and implications for education’ by the Australian Network for Quality Digital Education and Australian Centre for Social Justice and Inclusion. The report’s authors, Jason Lodge and Leslie Loble, write that;

“The true educational risk of AI is not simply that students will use it to cheat on an essay. The far more profound risk is that AI may fundamentally interfere with the cognitive processes of knowledge construction and verification, the very processes that build the long term memory stores and subsequent skills upon which the majority of critical thinking depends.”

See also the Guardian’s report that pupils using AI are losing their capacity for critical thinking, according to a survey of secondary school teachers in England.

Sarah will report back on their experiences of AI in Iceland next week.

Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

State of AI
Stanford University’s 2026 AI Index report is out and includes a chapter on education. It’s predominately focused on the US but highlights that while more than 90% of countries now offer computer science to primary or secondary students, AI education has been slower to take hold – only China and the United Arab Emirates have mandated AI education, starting with the 2025-26 school year, signalling a shift towards formal AI instruction at the national level. There’s much more to explore in the report, but MIT Tech Review has a more general overview, focusing on the unprecedented speed of this technological gamechanger: the top models just keep getting better, people are adopting AI faster than they picked up the personal computer or the internet but “the benchmarks designed to measure AI, the policies meant to govern it, and the job market are struggling to keep up”.

China introduces AI anthropomorphism law
China has introduced a law on AI anthropomorphism and it’s the strictest to be found anywhere, writes Luiza Jarovsky. It’s received little attention but she argues that it offers a real-world example of a legal framework that acknowledges AI-related human vulnerabilities and proposes contextual technical measures to prevent AI-anthropomorphism-related harm. In relation to children specifically, she notes that, “among the prohibited anthropomorphic AI practices is generating content for minors that causes them to imitate unsafe behaviours, induces extreme emotions, or leads them to develop bad habits, which may affect their physical and mental health.”

AI literacies
This AI Literacies Framework for young adults aged 14-19, from the BBC and WAO, is aimed at public service media organisations but may have a wider interest for educators, drawing as it does on analysis of 40 AI literacy frameworks alongside expert interviews and survey data. It finds that existing resources prioritise technical and functional skills over critical evaluation and there is a risk that AI literacy becomes disconnected from other types of literacy (digital literacy, information literacy and media literacy), to which it relates and needs to be connected.

• Matthew Wemyss has drawn attention to an advert for a full-time AI literacy role in a Scottish school and highlights what this implies for all schools, especially with PISA test AI literacy assessments around the corner. He’s also got an interesting report on Europe Schoolnet’s Working Group on Artificial Intelligence in Education.

Khanmi no-go
In January the UK government announced plans to develop AI tutoring tools to provide one-to-one learning support for up to 450,000 disadvantaged pupils. Bridget Phillipson, the education secretary, said at the time: “AI tutoring tools have the potential to transform access to tailored support for young people, taking tutoring from a privilege of the lucky few, to every child who needs it – so all children can achieve and thrive.” However, in a blow for the hopes of AI tutoring initiatives, Sal Khan, founder of Khan Academy, has admitted the failure of Khanmigo, saying that AI has not allowed for the creation of an effective super-tutor. In this LinkedIn post, Will Van Reyk, director of Innovation at St Paul’s Girls’ School, suggests that “there was a flawed assumption in the original vision: that educational transformation in the age of AI is primarily a technology problem [rather than a ‘what is education for’ problem].”

Quick links

• Every day this week, BBC Radio Four is broadcasting a series about what it’s like to be a girl growing up in the UK today, and the documentary maker, Catherine Carr, has also written about her findings.

• Pair with this Guardian article about the masculinity crisis brewing in UK schools. Also, SWGfL has launched a resource to help schools and educators address issues relating to the manosphere; and the Fawcett Society has published a report on AI and gender equality.

• Hello World’s latest podcast shares actionable tips about how to teach children and young people about cybersecurity.

• Thanks to the Online Safety Act Network for the following shares:

• Sextortation attempts involving under-18s have risen by 34% in a year, according to the Report Remove service run by NSPCC and Childline.

• Parents should teach their children about online safety and data privacy as an essential skill, like road safety, says the ICO, with eight and nine year olds most at risk.

• There’s a call from SWGfL for professionals working with children and young people to take part in their survey of views on social media bans; and details of a young people’s town hall event, hosted by Flippgen and 5 Rights, to feed into the consultation with expressions of interest from teachers and youth leaders sought here.

• Meanwhile, Greece is the latest country to legislate for a ban, and the Guardian’s online games editor argues that banning teens from playing online games is a “terrible idea”.

We’re reading, listening, watching…

Are we making ourselves less intelligent?
While held up on the M25 last week, Michelle caught some of Radio 4’s Radical with Amol Rajan on the so-called ‘reading recession’. In it, writer James Marriott worries with Rajan that a less literate, more screen-dependent public may be more vulnerable to misinformation and less capable of meaningful democratic participation. It raised interesting questions but also seemed to suffer from overreach - are screens really to blame for everything? - so we were pleased to see Doug Belshaw’s post about this programme, which unpicks what he describes as the moral argument being made about reading, literacy and screens.

Technology traps
People were worried about tech dependency back in 1978, too, as this bleak BBC film shows. Only then it was about lifts breaking down…

You are a scamp!
A charming account of the labour of love that was the creation of Mr Chatterbox, a language model trained from scratch on a corpus of over 28,000 Victorian-era British texts published between 1837 and 1899, drawn from a dataset made available by the British Library.

We haven’t watched it yet, but Grayson Perry Has Seen the Future, a new series on Channel Four, is getting a lot of attention.

Give it a try

The Tenth Muse

This quirky art discovery engine contains more than 120,000 artworks from museums and institutions, searchable by feeling, mood, atmosphere, era, and medium.

Plus, for fun:

• How accurately can you recall and recreate a colour?

• How good’s your text abbreviations game?

Connected Learning is by Sarah Horrocks and Michelle Pauli

We’re heading straight to the news, views and resources this week, and then we’re taking a two-week break for the Easter holiday, back in your inboxes on 17 April.

Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

AI in HE: present or producing?
The HEPI student generative AI survey, now in its third year, is notable for cutting through the hype and speculation with clear data on how students are actually using AI in UK HE and how they feel about it. Top findings are that student use is near universal with 95% using AI in some way and 94% using it to help with assessed work but institutional adoption is lagging behind with only 36% feeling encouraged by their institution to do so, and 38% saying they are provided with AI tools. Interestingly, around 40% of students feel their use of AI affects their ‘loneliness’ but split into half who feel less lonely and half more lonely.

• Pair with new research from WonkHE suggesting that “Students are submitting work they cannot fully explain, facing AI policies that do not function in practice, and responding strategically to assessment systems that reward production over understanding”. It identifies an important divide between whether universities are asking students to “be present” (think, struggle, understand) or merely “produce”.

Scottish AI guidelines and guardrails
Scotland has just published Artificial Intelligence (AI) in schools: guidelines and guardrails, setting out core principles of the safe and effective use of AI in education, centred around the principle that, “the presence of AI must not, in any way, diminish human-centred education; it must only strengthen it.”

AI, family and community in early childhood education
What happens when you put generative AI in the hands of families and communities? Researchers from UC Irvine, Harvard and University of Michigan partnered with Latino families in California and Michigan to co-create culturally relevant e-books. Parents drafted stories, with the help of ChatGPT, about grandchildren video-chatting with abuelos in Mexico to learn about growing corn, and celebrating the differences between gorditas, arepas and pupusas. An artist then transformed AI-generated concept images into final illustrations. The project flips the script on AI in early learning – instead of replacing parents or reinforcing stereotypes, it amplifies marginalised voices and lets families bring their lived experiences into their children’s education. The researchers emphasise that this kind of work is best done in partnerships between families, educators, early learning researchers, artists and technology designers whose collective expertise leads to products that none of them could have made on their own.

Quick links

• The latest edition of the excellent (and free!) Hello World magazine is out and it’s focusing on digital safety, suggesting new ways to teach online safety and cyber security.

• This week’s social media ban story comes from the Guardian, which reports that 300 UK teens will take part in a six-week trial to test how different restrictions affect young people’s day-to-day lives

• The Commission into Countering Online Conspiracies in Schools has published 10 new findings in a report examining how the spread of conspiracy beliefs, misinformation and disinformation in classrooms.has evolved over the past 12 months across England. It finds that teachers say they lack training and confidence to deal with these difficult conversations and that it extends to primary schools, with younger children also being exposed to offensive and harmful material at an age when their ability to assess it critically is still developing. Schools Week has a good summary.

• Young people are consuming news less frequently than older people and are less interested in it, say the authors of this Reuters Institute report. Around 15% of 18 to 24 year olds are using AI to access news weekly compared with just 3% of those 55 and over. On social and video networks, young people say they pay more attention to individual news creators (51%) than to traditional news brands (39%), in contrast with those 55 and older, who say they pay more attention to the news media.

• Although the BBC is the world’s biggest commissioner of new content for children, it is spending less. When CBeebies was launched in 2002, the BBC spent 4% of its income on children. Now it is barely 2%.

• The subtitles myth: the popular assumption that watching TV with subtitles will help children to become better readers doesn’t hold true, write reading researchers Kathleen Rastle and Anastasiia Lopukhina in the TES

• A jury in Los Angeles found ‌Google and Meta liable for a woman’s social media addiction in a landmark social media lawsuit. It’s seen as a bellweather decision that will inform thousands more cases against social media firms for creating addictive algorithms. The Guardian calls it a big tech reckoning.

• The Information Commissioners’ Office has written an open letter to tech firms calling on them to strengthen age assurance measures so young children can’t access services that not designed for them. ICO has also written directly to platforms, starting with TikTok, Snapchat, Facebook, Instagram, YouTube and X.

We’re reading, listening, watching…

Are you an apocoloptimist?
The film AI Doc: Or How I Became An Apocaloptimist opens this week. The team behind this groundbreaking documentary are interviewed on Your Undivided Attention podcast. Told through the perspective of filmmaker Daniel Roher as he prepares to become a father, the film investigates the existential threats and potential promises of artificial intelligence. Roher conducts interviews with leading experts in the field to understand the world his child will inherit, balancing anxiety with apocaloptimism.

• Pair with Radio 4’s Artificial Human series, especially the recent episode on AI and the future of learning

Can you feel it?
It may initially seem as if this article by Benn Stancil, fascinating read as it is, has little to do with education. It contrasts Anthropic’s huge study on what people think about AI, with a World Bank project, Voices of Poverty from the 1990s:

“The final tally: The World Bank – 60,000 interviews, from 23 countries, tens of thousands of hours of work done by almost 400 people, taking two years from start to finish. Anthropic – 81,000 interviews across 159 countries and 79 languages, in three months, done by 25 people. It is a breathtaking acceleration; a staggeringly short ride in an incomprehensibly fast machine. Think of what is possible now. Think of what we can learn.”

But Benn goes beyond the numbers to look at what is gained, and what is lost, from working at this scale – “if we don’t feel what people say, do we understand it? If we don’t grapple with it, does it change us?” and relates it to learning and the value of grappling with text, with concepts.

Give it a try

Museo

Already used every ‘coffee cup next to laptop’ image on Unsplash? Museo is a visual search engine that connects you with images from the Art Institute of Chicago, the Rijksmuseum, the Harvard Art Museums, the Minneapolis Institute of Art, the Cleveland Museum of Art, and the New York Public Library Digital Collection. Most are free to use.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Last week Sarah presented a keynote and workshop at the Bromley Headteachers’ Conference, focusing on digital strategy and current challenges presented by AI in education. She explored digital transformation in schools as a strategic and cultural process, where the use of digital technology is a critical thread across all aspects of a school’s operation from finance and HR to communication, curriculum and safeguarding.

The first session emphasised asking critical questions about technology purchases, investing in staff skills as much as equipment, and being wary of personal enthusiasms or restrictive contracts that lock-in the school. Digital technology in the curriculum is fundamentally an inclusion issue and being digitally literate is an entitlement. With up to 19 million people in the UK in digital poverty, and a significant portion of the population lacking basic digital skills, addressing this requires a renewed emphasis on digital literacy in the curriculum and ensuring all staff use accessibility features effectively to meet the needs of all learners, including SEND pupils.

Sarah and the headteachers then considered different aspects of AI in schools, such as AI policy, acceptable use, tool approval processes, data protection, and transparency of risk. They shared AI productivity examples, such as creating automated cover rotas, using AI to analyse mock exam data, setting reminders and saving time on writing up meeting minutes. They also discussed the challenges of mapping AI literacy into different curriculum areas – computing, RSE, PSHE, English, maths – managing AI-generated parent complaints, cognitive offloading (using AI to extend thinking) and cognitive outsourcing (letting AI substitute for thinking) in headteachers’ own and students’ use of AI.

Meanwhile, Michelle has been filming videos for the second phase of the DfE safe and effective use of AI resources for school staff, with a particular focus on SEND and business and operations. Expect to see the refreshed toolkits, including more interviews with insights from practitioners, later this spring.

AI roundup

Human-made is becoming a valuable badge
The BBC reported that declarations such as “Proudly Human”, “Human-made”, ‘”No AI” and “AI-free” are appearing across films, marketing, books and websites. Organisations across the world are racing to develop a universally recognised label for “human-made” products and services as part of the growing backlash against AI use.

Teachers should have to earn the right to use AI
Richard Watts, a teacher of business at Richard Taunton Sixth Form College in Southampton writes in The TES this week that AI use must not come at the expense of hard-earned professional wisdom.

“Lessons may appear slick, resources may look well-structured and workload may feel lighter. But the teacher may be doing less of the slow thinking that builds long-term expertise: analysing why a task fell flat, spotting subtle misconceptions or deciding when to let students struggle productively….This is why AI should be understood as something that must be earned before it can be used to augment a teacher’s work.”

• And, what’s more, should all classrooms have an aviation-style challenge structure and routine - for teachers and students alike?

How to teach about AI – unplugged
We’re fans of ‘unplugged’ approaches to teaching technical concepts (at the CLC we taught computational thinking in early years with pattern spotting in playgrounds and using action songs such as Heads, shoulders, knees and toes as a way of introducing algorithms) and in this Raspberry Pi blog post Jane Waite outlines an intriguing technique for teaching about neural networks using ropes (actual ropes the students hold). The case study she presents from Salomey Afua Addo, a researcher at the University of Cambridge, is based on farming and crop diseases as a familiar problem context for Ghanaian teachers and their students, but it could be adapted for other contexts. Sets of students act out the processes of training a neural network, including forward propagation, evaluation and backpropagation.

• There are more practice-based insights in these interviews Michelle produced for the DfE, published this week on the DfE blog. Alan Stupple, director of maths at Bedfordshire Schools Trust, shares how he’s using the DfE support materials on AI across schools, and Georgia Aspinox, lecturer in primary mathematics education at the University of Reading’s Institute of Education, describes the challenges in teaching trainee teachers to use AI.

Quick links

• Last July the DfE unveiled its proposed changes to the RSHE guidance. Ministers ditched the previous government’s plan for age limits on certain topics and a section forbidding teaching about gender identity. The changes mean primary schools will be encouraged to teach about same-sex parents, and there will be a new emphasis on misogyny and sexual violence, deepfakes and sextortion. But schools have been left in the dark over when the training grants will be made available to help them implement the new RSHE guidance from September, reports Schools Week.

• Adding some nuance to the social media ban debate: algorithm-driven and influencer content-stuffed apps (Instagram, X) appear to be worse for mental health than those focused more on connection (Whatsapp).

• A new book by David Buckingham (Honorary Professor, UCL Institute of Education) is definitely on our reading list: The End of Information addresses whether we are really living in a ‘post-truth age’, whether ideas like information, knowledge and truth have passed their sell-by date, and if ‘information disorder’ is leading to the terminal decline of democracy. The book considers the implications for education in relation to key issues like literacy and knowledge.

• Pair Buckingham’s book with Media Literacy for the Communication Ecosystem by Julian McDougall, which makes the case for media literacy as a human right and a foundation for civic resilience.

• The latest edition of Jisc’s excellent digital sustainability roundup from Cal Innes is out and is heavy on data centre stories. It also features a paper documenting how tech companies are using interface design to push users towards generative AI features whether they want them or not, a report offering one of the first standardised frameworks for assessing the environmental impact of AI systems, and an article asking whether asking what the energy cost is of an individual AI query is even the right question.

• A Mac Mini inside a classic LEGO computer.

We’re reading, listening, watching…

Is a social media ban for under-16s the right move?
As the government considers whether a social media ban for under-16s would be the right way forward, Parent Zone raises several important questions that also need answering. The piece reports that leading academics have outlined serious reasons for not rushing to ban social media and the critical importance of waiting to see what the effects of the ban in Australia are first. Parent Zone is asking:

1. How do we reassure parents that change isn’t just coming – but that it has started?

2. When will we stop ignoring parents (until they get so angry the crescendo of frustration reaches MPs and we end up with demand for a solution most experts know is unlikely to work)?

3. How quickly can we move to make sure the smaller, most harmful sites are category 1 services (and therefore required to follow all of the requirements of the act)?

4. In an age of AI, with wearable technology coming onto the market at speed, how do we avoid heading down a rabbit hole?

5. How can we ensure that the debate is well informed by technology experts in a field that has become so polarised and so contested?

6. Are we brave enough to make the sort of serious changes that will give parents what they need in a way that is futureproof and meaningful?

7. Finally, are we patient enough to wait for the evidence?

Pair with the latest newsletter from the Online Safety Act Network.

The Race to Control the World
In this new BBC Radio 4 series, Misha Glenny, journalist, author and expert on global organised crime and cybersecurity, explores some of the key areas of the global tech battle - from AI to robotics to satellites and explains how it’s going to impact all our lives.

Give it a try

Can you predict what AI will do?
How good are you at predicting how well Claude, ChatGPT and Gemini will perform on simple tasks? Adam Kucharski set AI tools some simple tasks and gave them 10 attempts at each one. See if you can predict the results – and then how you compare with others.

Connected Learning is by Sarah Horrocks and Michelle Pauli

We’re heading straight to the news, views and resources with a shorter newsletter this week. Normal service to resume next week. Probably. Don’t forget, if you’d like to share an account from your own practice, your thoughts on digital developments in education or any tips or feedback – please drop us a line on email or catch us on LinkedIn.

AI roundup

EU guidelines on ethical use of AI in education
The European Commission has published guidelines on the ethical use of artificial intelligence and data in teaching and learning. It includes questions and scenarios with practical tools and examples for classroom and school settings and more principles and legal context, including clear explanations of the AI Act, General Data Protection Regulation (GDPR), and the principles of responsible AI use.

What do you notice about your day when you listen to it?
Civics of Technology has created an AI literacy lesson plan to help children compare human and AI soundscapes of their local community through creating sound journals. The lessons encourage pupils to ask: How do we understand ourselves and our community through sound? How does AI describe our community through sound? What do we notice about our community when AI describes its sounds?

Ofqual boss hears pupils’ confessions on AI
Ian Bauckham, chief executive of Ofqual has revealed that pupils have told him about how they use AI in academic work without their teachers’ knowledge, reports Schools Week. He has called upon examination boards to implement stricter measures against cheating. Bauckham highlighted that, while detection rates are low, there’s considerable anxiety among school leaders regarding the real prevalence of AI-related academic dishonesty.

Otterly interesting
Ethan Mollick’s back with his latest otter on a plane test and he’s using it to show just how far AI has come in the past three years and a hint of the impact it may yet have due to practical agents, jagged exponential improvement and the ability to radically experiment with the nature of work. It’s a “rolling and unpredictable environment” he warns, with the feeling of uncertainty that’s only likely to increase However, he strikes a note of reassurance amid the unpredictability:

“But uncertainty is not the same as helplessness. When a technology is this powerful and this unsettled, the choices that individuals and organisations make right now matter more. We can see the shape of the Thing now, but we can still influence the Thing itself, and what it means for all of us.”

Then:

Now:

US AI literacy framework
The US Department of Labor has published an AI Literacy Framework to guide workforce and educational preparation efforts. Al Kingsley has written a guide for schools to the framework.

Joseph Stiglitz says buckle up
“Joseph Stiglitz wants you to hold two ideas in your head at the same time,” writes Catherina Gioino in Fortune. “The first: an AI bubble is building; it will likely burst; it will hurt the macroeconomy; and workers will bear the cost of a displacement we have no institutions to manage. The second: survive that transition and the technology that threatens your job today may end up becoming your most useful coworker.” That’s a big if there.

AI’s ‘women problem’
In this week’s Slow AI newsletter, Sam Illingworth describes why AI systems have a gender problem and describes how,

“The models learned our biases and turned them into defaults…The absence of women in the room is a design flaw. It produces systems that do not work for half the population. You would not ship software that crashed on half of all devices. Yet we ship AI that fails on half of all humans and nobody blinks.”

It’s an issue / catastrophe that those of us familiar with Caroline Criado Perez’s excellent work will already recognise across a range of fields, from medical research to car safety. UNESCO’s 2024 study on gender bias in large language models found that Meta’s Llama 2 described women as working in domestic roles four times more often than men. Across the models studied, female names were associated with ‘home’, ‘family’, and ‘children’. Male names were associated with ‘business’, ‘executive’, and ‘salary’. Open-source models assigned women roles including ‘domestic servant’, ‘cook’, and ‘prostitute’. Men got ‘engineer’, ‘doctor’, and ‘teacher’.

• Pair with Victoria Hedlund’s paper on gender bias: Gender equity in GenAI science explanations

Quick links

• MPs have rejected an Australia-style ban on social media for under-16s, and have instead backed flexible ministerial powers. Record your views via the government consultation - growing up in an online world.

• Meanwhile, some early data suggests that the Australian ban has had a minimal effect on the number of teens on social media in Australia.

• Parent Zone’s position is that a social media ban is not the correct road to go down. In its Tech Shock podcast this week, Parent Zone’s CEO Vicki Shotbolt discusses whether the Online Safety Act has failed and what are its big gaps.

• In Safeguarding in schools: What leaders and DSLs need to know for 2026 a practical webinar. Victoria Hedlund, Andrew Bernard, Sallie Greenhalgh and Katherine Cauchi look at what safeguarding in 2026 requires in practice, explore the realities of safeguarding leadership and the DSL role in Trust and school contexts; unpick misogyny and harmful behaviours, and what schools can do that actually helps; make sense of AI-related safeguarding risks, including deepfakes, bias, and ethical challenges.

• A research paper; Computing reform: the exodus of ICT teachers by Amy Box and Sarmin Hossain, looks at how the introduction of the new national curriculum’s computing programmes of study in the 2010s, with their emphasis on computer science, which required secondary school ICT teachers to transition to teaching Computing, has led to teachers leaving the profession. The research indicates that these teachers felt inadequately supported throughout the reform process. This significant educational loss has reduced diversity within computing classrooms.

• The DfE has announced that the young people will be able to take the very first V Levels in education, finance, and digital in 2027, as part of reforms to post-16 system that will prepare them for future jobs.

• Abuse can happen anywhere – and it’s not always easy to recognise. Take the Enough Campaign quiz to see how well you can spot what crosses the line. Pair with this Today In Focus podcast which features interviews with Abbey, 15, and Josh, 16, about their experiences of social media, the manosphere, misogyny and the growing divide between boys and girls.

We’re reading, listening, watching…

Teacher v chatbot
First a mea culpa. Last week we featured a long read from Peter C Baker, a freelance writer and novelist in the US, who retrained as an English teacher and has chronicled his exploration of what AI means for English teaching. We ended the piece by suggesting that:

“Ultimately he advocates a measured blend that preserves AI‑free spaces for deep reading, writing and discussion, while harnessing AI’s feedback capabilities under clear instructional guidelines.”

Peter has been in touch to say that “just to clarify – my article does *not* end with advocacy for AI feedback tools in schools.” Apologies, and do read Peter’s excellent piece.

• Pair with Dan Fitzpatrick’s podcast about the article.

Is this the Broligarchy’s first world war?
In this powerful and disturbing long read, Carole Cadwalladr connects and exposes the alliance of Trump, Silicon Valley and a global axis of autocracy: a tech bro oligarchy, aka the Broligarchy.

“The same logic that built recommendation engines and social media algorithms is now being applied to the industrial generation of kill lists. Chatbot systems that entered public life writing emails, summarising documents and telling you what to have for dinner now help determine who lives and who dies.”

How to work with me
What does your AI tool need to know about you? How about creating a user manual for it? Doug Belshaw explores how to do it and whether it’s an “advanced form of digital literacies” in this short post.

Give it a try

Moebio

We’ve been looking at Moebio this week, thinking about how to visualise and teach about “dimensions” in LLMs; the complex mathematical and statistical features an AI uses to understand words and the relationships between them. The concept of “dimensions” in AI pattern recognition, within LLMs is a core technical aspect of how these systems understand and process language to predict the next most probable word in a sequence, creating coherent, human-like text. CLick on the question mark in the top right of the black pane for more explanation of this ‘look into a machine’s mind”.

Connected Learning is by Sarah Horrocks and Michelle Pauli

An ed-tech tragedy? Educational technologies and school closures in the time of Covid-19.

This interview on the Education, Technology, Society podcast resonated for Sarah, bringing back experiences of supporting schools in the UK and in Jordan with remote learning during the Covid-19 pandemic and devising the BlendEd initiative; a programme of professional development in blended learning pedagogy commissioned by IBM UK Charitable Trust. Sarah and the CLC also took part in Co-Learn, an Erasmus project looking at blended learning pedagogy and delivery during and after the pandemic in Sweden, Denmark, Finland, the UK and the Netherlands. Teachers and headteachers explored themes such as pace, place and path, how student-teacher feedback and dialogue can be supported by technology, how technology can enhance peer-to-peer feedback, and whether digital communities of practice can enhance student learning.

In his book An Ed-Tech tragedy? Educational technologies and school closures in the time of COVID-19, Mark West, from UNESCO argues that we should look back on Covid remote schooling as an edtech tragedy and use our pandemic experiences to develop radically different visions of digital education. Mark’s book (which is free to access) describes how the pandemic fundamentally changed our dependency on technology platforms and products, with billions of dollars rushed into purchasing edtech infrastructure that schools are now stuck with. He also argues that the Covid experience provides us with valuable lessons about the frailties and unfairnesses of edtech.

Listening to Mark and reading extracts from his book made Sarah reflect on the success of the UNICEF Learning Bridges programme which she and her Connected Learning Centre colleagues contributed to. What distinguished this ambitious initiative to teach over 600,000 children in Jordan remotely was that Learning Bridges was designed as a programme that could be fully paper-based, with QR codes providing links to additional resources provided online. This was intended to overcome the lack of technology faced by low-income families and families living in remote areas.

In his book, West shows that in contrast, much of the global evidence reveals a sombre picture of how increasing dependence on edtech led to “unchecked exclusion, staggering inequality, inadvertent harm and the elevation of learning models that put machines and profit before people.”

As we face yet more global crises, West’s conclusions are salient:

“Crises that necessitate the prolonged closure of schools and demand heavy or total reliance on technology have been exceedingly rare historically. Future crises may present entirely different challenges. The trauma of the pandemic has, in many circles, functioned to elevate technology as an almost singular solution to assure educational resilience by providing flexibility in times of disruption. Investments to protect education wrongly shifted away from people and towards machines, digital connections and platforms. This elevation of the technical over the human is contradictory to education’s aim to further human development and cultivate humanistic values. It is human capacity, rather than technological capacity, that is central to ensuring greater resilience of education systems to withstand shocks and manage crises.”

AI roundup

Britons and tech
There’s an urban-rural divide in Britain across views of the merits of new technology, according to new research from More in Common. However, Britons are far more likely to say that technology makes them feel connected to others, rather than lonelier, especially among the younger generations. There’s an even stronger divide when it comes to AI, with Gen Z twice as likely as the rest of the public to use AI to learn, create, write, or even provide emotional support. Older generations (Gen X (!) and up), meanwhile, say that they use AI sparingly if at all. Of all the roles respondents would least like to see replaced with AI, teachers win resoundingly, with 69% saying that we should never explore or consider AI chatbots instead of teachers in the classroom. Protecting children from harm is the top priority for government in relation to technology:

Notebook LM data warning
Matthew Wemyss has shared an important post explaining why it’s not OK to upload an EHCP to NotebookLM to get a quick summary. Even though Google states that NotebookLM will not use uploaded documents for training its AI models, and so it seems to be safe, “not used for training” and “handled in compliance with GDPR” are two very different things. EHCPs contain Special Category Data under GDPR.This creates two serious compliance problems: data leaves your setting’s controlled environment, and you lose organisational oversight. He concludes, “Even if your school has Google Workspace for Education, uploading an EHCP to NotebookLM is a compliance failure waiting to happen. There is no lawful basis that comfortably covers processing Special Category Data about children through a tool where you cannot control data residency or demonstrate adequate governance. The 30 seconds you save isn’t worth the risk to a child’s data, or your school’s reputation.”

• Meanwhile, Philippa Hardman has a NotebookLM for learning design 101 with five evidence-based methods to try.

How to avoid cognitive stunting
We regularly feature the work of Rebecca Winthrop, who leads the Center for Universal Education at the Brookings Institution, and recently published a report called A New Direction for Students in an AI World. She and her colleagues conducted an extensive ‘pre-mortem’ of AI in the classroom, speaking with hundreds of educators, students, policymakers and technologists worldwide. In this Undivided Attention podcast from The Center for Humane Technology, Rebecca shares her findings in a really practical way based on real classrooms and conversations with young people. She says that the way both teachers and children are using AI is eroding relational trust between them. Her findings show that narrow strategic use of AI with vetted content can be beneficial but that open unstructured use of AI can impair children’s cognitive ability for independent learning and thinking and to take critical feedback. She calls it ‘cognitive stunting’. She calls for 1) shifting what teaching and learning looks like so it’s not hackable by AI, 2) preparing all staff in schools and education systems (as well as parents) to understand how to use AI well 3) have clear regulation and guard rails established in schools and organisations overseeing schools.

Time to swap out ChatGPT?
Historian Rutger Bregman thinks so and urges a boycott to send a message to Silicon Valley. He notes, “I want to be clear: I am not anti-AI. I use AI tools in my work every day. This is not about rejecting technology. It is about rejecting the idea that we have no choice but to fund a company that’s bankrolling authoritarianism.”

Learning to program with LLM tools at KS5
Raspberry Pi is launching a new research study for Key Stage 5 (A-level/Scottish Highers) Computer Science teachers in England, Scotland and Wales. It aims to develop a teaching framework and resources for using LLM tools in programming, covering how they work, safe and effective use, and helping students critically evaluate AI outputs. The study involves two in-person workshops in 2026 and 2027, and teachers can apply via this form.

Quick links

• Teaching students how social media algorithms work can change how they engage with what they’re shown, writes Tracey Neale, assistant headteacher at Ysgol Gyfun Cwm Rhymni in Caerphilly, Wales. She references the Safer Scrolling report by UCL and Kent University and argues that it’s essential to include boys as part of discussions regarding online misogyny, as hardline approaches can entrench negative ideologies and ways of thinking.

• Another reminder that Molly v the Machines is now available to stream on Channel 4 (see trailer).

• A youth-led debate organised by Durham Youth Council on whether social media should be banned for young people takes place on Tuesday 17 March 10:00am, online.

• The UK government plans to introduce a digital complaints system to improve how schools handle parental concerns and strengthen relationships between parents and school leaders. ParentKind reports that more than five million complaints were submitted by parents in the last year alone. A Schools Week investigation revealed the scale of pressure this places on headteachers, with many losing sleep over the volume of complaints – a significant number of which are believed to have been generated using AI.

• England’s largest exam board has warned Ofqual that it risks “moving too slowly” on the rollout of digital exams, reports the TES.

• London Centric is a brilliant news resource investigating varied aspects of London from dodgy billionaire landlords and the risks of Lime bikes to snail farms in office blocks (yes, really). This week it’s turned its attention to the so-called ‘school wars’, spread through social media algorithms and built up into a social panic despite the Met Police saying they still haven’t encountered a single incident relating to the viral red vs blue school wars trend.

“Even some London teachers are quietly rolling their eyes. ‘It made me laugh a bit that the social media posts are like ‘bring compasses, a ruler, a metal comb’,’ said one. ‘My students can barely manage to equip themselves with a black biro. Plus, if we’re being real, everyone knows that ‘perilously sharp-edged broken protractor’ would be the true weapon of choice.’”

We’re reading, listening, watching…

The AI lottery
Last week we featured a great post from Ben Ansell which referenced Jorge Luis Borges’s story The Library of Babel in relation to AI. Now Doug Belshaw has picked up the Borges thread, speculating that the events the writer describes in The Lottery in Babylon are less speculative fiction and more 21-century reality when it comes to AI.

Teacher v chatbot: my journey into the classroom in the age of AI
Peter C Baker, a freelance writer and novelist in the US, retrained as an English teacher and has chronicled his exploration of what AI means for English teaching and his oscillation between AI‑rejection and AI‑enthusiasm. Ultimately he advocates a measured blend that preserves AI‑free spaces for deep reading, writing and discussion, while harnessing AI’s feedback capabilities under clear instructional guidelines. His model aims to retain the “friction” essential for learning while equipping students with the digital literacy they’ll need beyond school.

Understanding ‘good enough’ parenting and toddlers’ tech use
Janet Goodall, researcher and Professor of Education at Swansea University’s School of Social Science, discusses all aspects of parental engagement and her recent research into early years’ tech use on ParentZone’s Tech Shock podcast.

Give it a try

The Fish Doorbell
The Fish Doorbell is back! Help fish spawn in Utrecht by ringing the bell and alerting the lock keeper that fish are waiting to get through.

Connected Learning is by Sarah Horrocks and Michelle Pauli

Welcome back after our half-term week off, in Sweden (Sarah) and Amsterdam (Michelle).

This is our 100th newsletter! When we started in September 2023 we promised fortnightly “news, trends, insights, guidance and resources to help teachers and school leaders navigate the digital landscape, including the latest AI developments”. We quickly realised that the digital landscape was changing so fast that a weekly edition was needed to keep up.

In that time we’ve seen massive AI advances – we launched the newsletter just the week before the rollout of Co-Pilot and Bard. We’ve since charted AI’s evolution from a collaborative tool to an autonomous agent, witnessed all shades of opinion from optimism about AI integration to critical pushback and a concern for ethics and human rights, explored the smartphone debate, reported on guidance and examples of how to use AI in schools, reported the debate about social media escalating from school rules to national bans, covered the call for more digital literacy and less computer science in the curriculum in England, and considered concerns about all of our dependence on big tech.

We’ve also featured interviews with movers and shakers: Katarina Sperling, a researcher in pedagogical work at Linköping University in Sweden, Christian Turton edtech lead at Chiltern Learning Trust who led the project to create the UK government’s support materials for schools and colleges to use AI, Michel Buchner creative technologist and teacher in the Netherlands, Emma Darcy director of technology for learning at Denbigh High School. We’ve also published short essays on topics from the Jonathan Haidt’s Anxious Generation debate, extremism and young people, how AI is changing reading, writing and thinking, playful learning,the environmental impact of AI, film and media education, the Nordic digital pullback, to ’diversionalry soft-focus electronica’.

Over to you
Throughout we’ve kept the same format of rounding up AI news, quick links about all things digital and education, inspiring reads and listens and a tool or resource to try. Is it tme for a change? We love the feedback we’ve had from you over the time (whether comments on Substack and LinkedIn or emails), but would love to hear more about what you’d like to see in the newsletter over our next 100 issues – what would you like more or less of? Anybody you’d suggest as an interviewee? Perhaps you’d like to share your own digital education practice? Please let us know! Leave a comment on Substack or message on email or catch us on LinkedIn.

AI roundup

A guide to which AI to use in the agentic era
We highlighted Ethan Mollick’s One Useful Thing newsletter in our very first edition and, 100 newsletters on (did we mention that this is our centenary edition?), we’re still finding it to be thoughtful and, indeed, useful. He’s updated his guide to which AI to use and this eighth version represents a “very large break with the past” due to the rise of agentic AI. In this new guide he makes a helpful distinction between models (the underlying AI brains), apps (the products you use to talk to a model) and harnesses (they “let the power of AI models do real work, like a horse harness takes the raw power of the horse and lets it pull a cart or plough”).

AI to support deaf students
Jisc has been exploring how AI can support D/deaf and hard of hearing students and this is a helpful guide that covers not only the more common use cases (captioning) but also some that are more early stage and exploratory (AI sign language translation, smart glasses for live captions).

Safeguarding approach to GenAI
London Grid for Learning (LGfL) has collated its resources into a guide to help school leaders answer the question: How is your school ensuring there is a robust safeguarding-first approach to using generative AI? There’s a webinar, explainer video, posters, staff questionnaire and infographics.

Mixed methods come out top
A mixed picture emerges from this study on the effects of LLM use and note-taking on reading comprehension and memory, the result of a randomised experiment with 450 students aged 14-15 in seven schools in the UK. The research finds that, while note-taking promotes cognitive engagement and long-term comprehension and retention, LLMs may facilitate initial understanding and student interest. The study shows the benefits of combining AI use with traditional learning over using AI alone, and the AI skills that students need to maximise those benefits.

AI-generated feedback - when it works, and when it doesn’t
Philippa Hardman examines new MIT research that gave students identical AI-generated feedback – but told half it came from their human teaching assistant. Both groups found it equally helpful, but students who believed a person had read their work spent significantly more time on later modules and wrote more code. She suggests that the lesson is that technical success isn’t pedagogical success: learners need to feel seen.

AI;DR
Riffing off the established t;ldr (too long; didn’t read - which usually precedes a summary), a new term is calling out AI slop so other people can avoid wasting their time or to just make fun of the person peddling it: “AI;DR,” or “ai;dr,” short for “AI, didn’t read.”

Quick links

• Tech Policy Press has created a Global Social Media Age Restriction Tracker to map and monitor legislative efforts to control children and teens’ access to social media across the world.

• A couple of weeks ago we reported on the trend for breaking up with Big Tech and highlighted some of the alternatives on offer, from search engines to email – and now the Guardian has followed suit!

• The Information Commissioner’s Office, the UK’s data watchdog, has fined Reddit £14.47 million for unlawfully processing the data of young children after it failed to implement robust age assurance mechanisms.

• While we were away the DfE published a new draft Keeping children safe in education 2026 and Schools Week published everything you need to know about it.

• Raspberry Pi is offering a free short online course on using the PRIMM framework (Predict, Run, Investigate, Modify, Make) to teach programming.

• Mumsnet has launched a national advertising campaign calling for a ban on social media for under-16s. It features cigarette packet-style health warnings.

We’re reading, listening, watching…

The Apprentice’s Sorcerer
We know there’s currently no shortage of wide-eyed posts out there detailing the wonders of Claude Code but this one, from political scientist Ben Ansell, is a notch above. In it he describes a variety of experiments he set Claude to work on and the results, followed by more general commentary on what it means when “the brakes come off”. As he says, “I do think most social scientists are going to have to either figure out how it works or refrain from doing so but make peace with its existence. This is the ‘killer app’ and its precise method of killing you will be social scientists boring you to death talking about it.”

• But, if that’s made you feel anxious about not keeping up, here’s some words of wisdom from Oliver Burkeman. He claims that “the reason ‘you’re not ready for what’s coming next’, is that we’re never ready for what’s coming next.”

What the left is missing on AI
Dislike AI all you want but don’t ignore it, says Dan Kagan-Kans in this Transformer post that’s thrown the cat among the pigeons. He accuses the left of negligence in leaving the debate on the threats and opportunities of AI to the right because key commentators and academics refuse to engage seriously with the arguments and persist in dismissing AI as ‘word guessing machines’ or stochastic parrots. He argues that in 2023, when chatbots were more toy than tool, AI-as-autocomplete was maybe a defensible position. But now? There are many reasons for AI scepticism but to dismiss the technology out of hand carries its own significant risks. Read a rebuttal from Brian Merchant in Blood in the Machine: Actually, the left is winning the AI debate.

Give it a try

As we noted above, we’ve previously covered breaking up with Big Tech. Here are a couple of alternative LLMs that claim to offer full privacy: Confer and Maple. Caveat, we haven’t yet tried them. (And sorry, it’s not quite as fun as last week’s Gradient Horse…)

Connected Learning is by Sarah Horrocks and Michelle Pauli

Housekeeping: we’re off next week for half term, back in your inboxes on 27 February.

The chatbot v human creativity challenge

Last month, Sarah guest-lectured for UCL’s Creativity: from theory to practice in English, Media, Drama module for student groups studying MA in English and MA in Digital Media: Education. The focus was on whether creativity and creative thinking can be taught and assessed, and where AI fits into the picture.

As part of the UCL lecture the students looked at concepts of creativity and creative thinking. They examined the PISA creative thinking tests, explored references to creativity in different countries national curricula, considered whether the PISA tests are worthwhile and looked at how AI is changing how students learn to read, write and think. Sarah also shared examples of school students’ creative use of digital technology from the Connected Learning Centre and drew on her work with Sue Ellis and Myra Barrs on the Creative Learning Assessment and Hill Mead Observing Learning Record.

PISA creative thinking assessment

International OECD assessments are interested in creative thinking because of the economic imperative. The World Economic Forum’s 2025 Future of Jobs report ranked creative thinking as the fourth most valued core skill by employers but governments don’t necessarily value creativity in national curricula.

In the PISA standardised test taken by 15-year-old students internationally, PISA defines creative thinking as:

“the competence to engage productively in the generation, evaluation, and improvement of ideas that can result in original and effective solutions, advances in knowledge, and impactful expressions of imagination”.

The OECD’s PISA framework defines creative thinking across four domains: written expression, visual expression, social problem solving and scientific problem solving.

In Sarah’s UCL lecture, students were split into six groups to work on three PISA creativity tests: social problem solving, written expression and scientific expression. For each test, one group of the students worked on the test themselves and one group used a generative AI chatbot. The groups then exchanged their results and discussed how well humans and generative AI did on the tests and whether the tests are of value.

In written expression, students found the human-authored stories more relatable and emotionally compelling. The AI-generated content was faster and technically competent. Students noted it seemed to have improved since previous exercises but it still lacked human authenticity and originality.

In social problem solving, humans came out clearly ahead. When tackling a challenge like making a library wheelchair accessible, the human group drew on empathy, cultural understanding and lived experience in ways the AI simply couldn’t match. The solutions weren’t just technically sound; they understood the human dimensions of the problem.

In scientific problem solving, the picture became more nuanced. The AI group could access a broader knowledge base and provide detailed scientific references. But the human group grounded their solutions in personal and community experience, bringing a practicality and contextual awareness that pure knowledge doesn’t capture.

AI and human thinking

We discussed the pressing pedagogical question: how do we support creative thinking in a world where AI can generate ideas, solve problems and produce content quickly.

We drew on some of the work covered in this newsletter in recent weeks and months including:

• the work of Paul Kirschner and Natalie Wexler on the difference between cognitive offloading and cognitive outsourcing. Cognitive offloading means using AI as a tool to support and extend your own thinking, like using a calculator to handle computation while you focus on understanding the mathematical relationships at play; cognitive outsourcing means letting AI substitute for your thinking, abdicating the cognitive work to the machine. This distinction matters as excessive reliance on AI can diminish critical thinking skills. But when used thoughtfully, AI can act as a “thought partner” that amplifies creative capacity rather than replacing it. (See Sam Illingworth on AI as Socratic coach)

• a study by Michael Gerlich at SBS Swiss Business School has found that increased reliance on artificial intelligence (AI) tools is linked to diminished critical thinking abilities

• Viktoria Mileva (lecturer in Psychology University of Stirling) post on assessment and AI and the need to shift from “tell me what you know” to “show me how you think and your process” assessments, reflection journals or short viva-style defences that reveal process

• Ethan Mollick’s work on generating ideas.

The key educational challenge isn’t whether to use AI or not. It’s teaching students to understand and use LLMs in a way that increases their own critical and creative thinking without surrendering cognitive agency.

By 2029, PISA assessments will include an AI and media literacy component, acknowledging that students need to “engage, create with, manage, and design AI, while critically evaluating its benefits, risks, and ethical implications”. Young people need to have the ability not just to be creative, but to understand how to think creatively and evaluate information in a world saturated with AI-generated content.

• See Empowering Learners for the Age of AI An AI Literacy Framework for Primary and Secondary Education OECD, European Commision, Code.org

AI roundup

Evidence, not hype
For Safer Internet Day 2026, EU Kids Online, a multinational research network, is launching its European survey of 25,000+ children aged 9-16 across Europe. This LSE blog post has a summary of the key findings, including that most European children are using generative AI, with use rising sharply with age, and mostly for practical use (homework) with a bit of play thrown in. Their sense of trust in AI is complex: “In sum, children’s decisions about AI are often deliberate and reflective, but also ambivalent, as well as playful and curious.” The authors call for policymakers to consider children’s rights, including their voices and best interests, as AI becomes embedded in everyday systems.

Anthropic AI safety researcher quits, warning of a world ‘in peril’
Mrinank Sharma, who led Anthropic’s safeguards research team, has resigned, suggesting the company has set its values aside. In a study published last week, reported in Forbes, Sharma found that interactions with AI chatbots may cause users to develop a distorted perception of reality. Sharma identified thousands of such interactions occurring daily and concluded that the findings highlight an urgent need for AI systems explicitly designed to support human autonomy and flourishing.

Effective prompting and Teams tools
Jisc’s AI in FE community is a gem with nuggets of good practice and hints and tips that apply well beyond FE – and it’s all accessible thanks to their habit of recording sessions and uploading to YouTube. The latest one features Rob Wraith, Head of Learning Technology and Digital Learning at Newcastle College Group, talking about using the ‘GCSE’ model (not what you think it is) for prompting, plus useful AI tools in Teams that people often miss

AI Is eating itself?
Rose Luckin cautions that anyone working in schools would do well to notice the economic warning signs flashing in AI financial markets. She highlights the Financial Times reporting this week that the US software sector is down 16%, with analysts describing AI as “eating itself”; automating the very data and analytics services that underpinned its own investment case. Luckin notes that as well as generative AI’s economics remaining precarious with billions in losses and business models that don’t yet stack up, the implementation of AI has consistently underdelivered. The implication for education is relevant, says Luckin:

“Schools cannot shed vendors, exit procurement contracts, or retrain staff at the same speed. But the underlying question is the same: is your AI adoption creating genuine value, or are you simply exposed to a technology whose costs and limitations you have not fully reckoned with? Adopting AI is not a strategy. Understanding where it adds value, and where it does not, is. The technology is eating its own supply chain. The question for education is whether we are building the human capacity to navigate what comes next.”

Quick links

• The world’s largest social media companies have been accused of creating “addiction machines” as a landmark trial began in California examining the mental health effects of Instagram and YouTube reports the BBC.

• Hackers urge Manchester pupils to use web skills for good.

• DfE call for evidence to inform the development of new parental guidance on screen time and usage for early years (0 to 5-year-old) children. And Susie Dent has some magical words.

• Staff in English secondary schools spend 100 hours a week enforcing smartphone policies.

• Research from Internet Matters’ Digital Wellbeing Index shows that gender gaps are narrowing around how girls and boys spend their time online but girls remain significantly more likely to encounter online abuse and harassment and, despite rising levels of distress among boys, girls continue to experience more emotional distress overall.

  

• The PSHE Association has produced a set of lesson plans and materials on understanding deepfakes and the law around them. (h/t More than Robots)

• UK charity Childnet has been accused of censoring young people it invited to speak about the biggest issues facing young people online as part of Safer Internet Day 2024.

• This academic study, Between guidelines and guilt: Parenting through the chaos of screen-time discourse, argues for reframing of digital parenting as a relational practice embedded within sociotechnical systems, rather than a purely private moral challenge (another h/t More than Robots).

We’re reading, listening, watching…

Can AI be your friend?
This longer read from Matthew Wemyss takes as its starting point his school’s Key Stage 3 debate on the motion “This house believes that you can trust an AI to be your best friend.” He’d expected a lively debate but there was unanimity - every single child concluded that AI can’t be trusted as a best friend. Wemyss examines their arguments and concludes that:

“what really got me was the range of their thinking. They talked about vulnerability. About why arguments between friends matter. About water usage in data centres. About subscription models. About specific lawsuits. About what it means for someone to choose you. One of them quoted OpenAI’s own statistics. Another did back-of-the-envelope calculations on water consumption per message.

“These are not kids who are afraid of AI. They use it all the time. They understand what it can do, and they’ve thought carefully about what it shouldn’t be asked to do. They’re drawing a line, and it’s the right one.”

We’re not in 2025 any more
Another LinkedIn longer read, this time from Dan Fitzpatrick, in which he examines the thin line between efficiency and autonomy in schools when it comes to agentic AI, and the need to outsource the doing, not the thinking. But he also highlights the AI adoption gap in schools – some schools have teachers using AI tools daily, others are bogged down in bans – and that gap compounds quickly because, he says, you can’t install culture like software. He urges school leaders to give teachers the permission and time to explore: “An AI sandbox, somewhere teachers could tinker safely without worrying about looking stupid. A working group of willing volunteers, some schools call them their AI Mavericks, testing ideas and reporting back to staff over lunch. Coffee chats where informal, low-pressure conversations surface about what teachers actually need, rather than what senior leadership assumes they need.”

AI ethics in education
In the Education Technology Society podcast, Neil Selwyn talks to Michał Wieczorek (University College Dublin) about how to think about tech ethics in a philosophically grounded manner.

Give it a try

Gradient Horse
Draw a horse, watch it run. Go on, give it a try! Click the ‘?’ for extra fun.

Connected Learning is by Sarah Horrocks and Michelle Pauli