Matthew Wemyss is a school leader, computer science teacher, and AI governance professional who runs IN&ED. From a career that began in media and IT to nearly 17 years teaching computer science and internet safety, his work sits where school leadership, safeguarding, curriculum, and technology meet. That combination makes him unusually well placed to answer a question many schools are wrestling with: how do you guide responsible AI use when the tools, risks, and everyday behaviour are changing faster than policy can keep up? In this conversation, Matt makes the case for governance that starts with reality. Staff are already experimenting. Students are already using AI. Parents are trying to make sense of it at home. A useful school response has to guide all three groups—not pretend the tools can be kept outside the gate. What You’ll Learn * Why an AI policy can fail when its rules do not match how staff actually work * The practical one-page guidance Matt recommends for schools that feel behind * Why AI governance needs voices from EYFS through to A level * How parent education fits into safeguarding and responsible AI use * A “human sandwich” approach that keeps thinking before and after the tool * How more than 3,000 students used design thinking before vibe coding in the Create Code Change competition * Why a teacher who builds a student-facing AI app may accidentally turn the school into an edtech provider * How schools can guide staff, students, and parents without waiting for a perfect policy A Reality-First AI Governance Starter Matt’s advice is deliberately practical. Schools do not need to begin with a perfect, exhaustive policy. They need a clear starting point that reflects what is actually happening. 1. Find out where you are Ask which AI tools staff are already using and what they use them for. Treat this as discovery, not a disciplinary exercise. Hidden use makes responsible guidance harder. 2. Publish a one-page guide Name the approved tools. State the non-negotiables clearly: no student data, personal information, confidential documents, or school policies in unapproved systems. Then give staff useful examples of what they can do. 3. Support the ecosystem you already have Whether your school uses Microsoft or Google, help staff get genuine value from the approved tools. If the supported option does not meet their needs, people will drift towards personal accounts and unsanctioned platforms. 4. Build a representative AI group Include colleagues from different phases and roles. AI use in EYFS may be almost entirely teacher-facing; at A level, students may be working directly with chatbots. Governance has to account for both. 5. Guide the whole community Staff need boundaries and practical training. Students need to learn when AI is useful and when it replaces thinking. Parents need clear information about age limits, safeguards, and productive use at home. 6. Review what changes AI guidance cannot be a document written once and filed away. Revisit the tools, risks, and behaviour regularly so policy continues to match practice. The “AI sandwhich” Matt describes the human role as the bread on either side of an “AI sandwich.” Students should arrive at a tool with context, a defined problem, and ideas of their own. After using it, they should return to human discussion, judgment, editing, or action. The Create Code Change competition put that principle into practice. More than 3,000 students from 19 countries used Canva Code to build an app connected to a Sustainable Development Goal. But the app was only part of the submission. Students also had to identify who they wanted to help, refine their idea, think through data ethics, and show their design process. Judges assessed the thinking as well as the finished product. That distinction matters. “Go and use AI” tends to produce generic work. A learner who brings a clear problem, knowledge of the user, and their own constraints can use the same tool with much more judgment. The Parent Piece Most Schools Are Missing School guidance only covers part of a child’s relationship with AI. At home, families are also navigating chatbot age limits, AI-generated search summaries, smart glasses, image and video generators, and tools with very different guardrails. Matt has run parent workshops since the early days of ChatGPT and is developing a parent handbook that explains the tools, productive uses, and safeguarding issues in plain language. His point is not that schools can control everything that happens at home. It is that repeated, practical communication gives parents a better chance of guiding it. Parent education is not an optional extra to an AI policy. It is part of the governance system. Vibe Coding Turns Schools into Edtech Vendors The episode also surfaces a less obvious risk. A teacher can now assemble a useful-looking app or chatbot in minutes. If that tool collects data or is placed in front of students, the school may no longer be simply using somebody else’s product. It may have become the provider responsible for security, data protection, and what happens when the system fails. That does not mean schools should suppress experimentation. It means enthusiasm needs an infrastructure check: What data does the tool collect? Where does it go? Who can access it? Are secrets or API keys exposed? Who is responsible if something goes wrong? Is building this ourselves actually the right solution? Schools hold medical information, SEND diagnoses, addresses, contact details, financial records, and other sensitive data. Responsible experimentation starts by treating that environment with the seriousness it deserves. Memorable Quotes “I think it’s got to match reality as well.” “The more context you can bring to these tools, the better.” “You could smash an app together in like 30 minutes, but everyone who uses your site’s data might be leaked in 30 seconds as well.” “If we’re to do anything, it’s got to be for a good reason and it’s got to be safe.” “If your staff are using it, guide them. If the students are using it, guide them. If parents need help, guide them.” Timestamps * 00:00 — Introducing Matthew Wemyss and responsible AI governance * 02:16 — How safeguarding concerns around AI are changing * 05:43 — What parents need to know about AI tools and age limits * 10:31 — Teaching internet safety through scenarios and reflection * 13:51 — Vibe coding without removing the thinking * 14:39 — Create Code Change: 3,000 students, 19 countries, and design before prompts * 18:19 — The human–AI sandwich: thinking before and after the tool * 20:39 — Using chatbots as coaches rather than answer machines * 22:32 — Why infrastructure, sovereignty, and local AI matter to schools * 23:13 — When a school becomes the provider of an AI tool * 27:11 — Why schools are high-risk data environments * 29:04 — The one-page starting point for schools that feel behind * 31:30 — Guide staff, students, and parents instead of ignoring AI use * 34:49 — Matt’s current projects and where to follow his work Resources & References Mentioned * IN&ED * AI Strategy Consulting for Schools * Open-Source AI for Schools: Building Ethical Data Infrastructure * Why Students Trust AI Less But Rely On It More * AI in Education: An Educator’s Handbook * AI Insights Focus: School Leaders, the EU Just Made AI Literacy Your Job * AI Insights Focus: A Manifesto for Responsible AI in Schools * Ctrl + Alt + Teach Where to Follow Matthew Wemyss * LinkedIn * IN&ED This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit braintoolsatschools.substack.com
