Hey {{ first_name | human }},
What a week it has been in the AI world. Grab a tea and a biscuit and get ready to digest.
TL;DR: The 60 Second briefing
⚡️NHS extends Copilot: The NHS is accelerating its roll out of Copilot to by providing 505,000 clinicians and support staff with access to the AI technology.
🧪Copilot Notebooks: Microsoft is bringing Copilot notebooks to their education customers.
🚨Fable & Mythos pulled: Anthropic were forced by the U.S. government to pull the latest releases of Claude from all markets apart from the U.S. due to security concerns.
📚 AI+education news
🧪 Copilot Notebooks > What it is: Microsoft has made Copilot Notebooks and Study Guide available to Copilot Chat users in education. Copilot Notebooks lets users collect notes, files and references in one AI-supported workspace. Study Guide can then turn those materials into learning aids such as summaries, key concepts, quizzes and revision support. Microsoft says this is available to Microsoft 365 Education A1, A3 and A5 users, with Study Guide available for education users aged 13+.
Why this matters: This is a sign that AI study tools are moving inside the platforms schools already use, rather than sitting as separate websites pupils need to visit. That matters for workload, safeguarding, access and governance.
Do this next: Use it first with teacher-controlled materials: a short knowledge organiser, a lesson slide deck, or a revision document. Ask pupils to compare the AI-generated study guide against the original source and mark anything missing, vague or misleading.
🌍 Wider AI updates
🚨 Fable pulled > What it is: Anthropic has suspended access to its advanced Claude Fable 5 and Mythos 5 models after a US government directive raised national security concerns. Reuters reports that the order limited access for foreign nationals, while Anthropic said the practical effect was that it had to disable the models for all users.
Why it matters: Powerful models are increasingly being treated as strategic infrastructure, not just software products. Access may depend on geopolitics, export controls, national security assessments and government confidence in model safeguards. For schools, avoid building critical workflows around one frontier model. While it is unlikely that any schools were using the latest Claude models, it may be worth considering an AI risk register to detail contingencies if models do become unavailable.
⚡️NHS extends Copilot > What it is: Microsoft says NHS England is rolling out Microsoft 365 Copilot to 505,000 clinicians and support staff. The move follows a trial across 30,000 NHS workers in 90 organisations, where staff reportedly saved an average of 43 minutes per day on administrative work or around five weeks per person per year.
Why this matters: This is one of the clearest examples of AI moving from “interesting pilot” to organisation-wide infrastructure. The key claim is not that AI replaces expertise, but that it may reduce admin load in systems where professional time is extremely scarce.
Do this next: For schools, the useful question is: which repeatable admin tasks are high-volume, low-risk, and still easy for a human to check? Start there. Meeting summaries, policy drafts, parent email first drafts, resource formatting and internal FAQs are more sensible starting points than pupil-facing judgement or safeguarding decisions.
Why this works: The NHS example shows the likely direction of travel: AI embedded inside everyday productivity tools, rather than used as a separate chatbot. That matters for schools because adoption often depends less on the cleverness of the model and more on workflow, training, governance and trust. The promised gains only matter if staff know when to use it, when not to use it, and how to check the output.
🎯AI concepts every teacher should know: 2. Tokenisation
Before AI responds to your prompt, it breaks the text into small pieces called tokens.
Tokens are not always full words. A useful analogy to consider is morphology.
The word ‘unhelpfully’ is a single word, but it is composed of four smaller parts, called morphemes: ‘un’, ‘help’, ‘ful’ and ‘ly’.
In AI terms, ‘unhelpfully’ is not processed as one single word, it is always processed as the four individual parts. This helps AI systems provide meaning
This is why AI really struggled with questions such as, ‘how many times does ‘r’ appear in strawberry?’ It does not see the individual letters, just the collection of tokens. It is important to note that morphology is a useful analogy for explaining tokenisation.
It does not mean that AI solely relies on morphemes. It is also uses characters such as spaces and punctuation to derive its tokens.
‘Till next week.
Mr A 🦾
Help a colleague save time by sharing this newsletter; distributing these ideas helps a friend get home on time and keeps our energy focused on what matters most: great teaching.
Safety & Privacy Notice
The tools and workflows mentioned are intended for professional productivity and educational enhancement. Users must ensure that any AI implementation remains compliant with their local data protection regulations and institutional safeguarding policies.
Data Privacy: Do not enter personally identifiable information (PII), sensitive student records, or confidential institutional data into public AI models.
Verification Required: AI-generated content can be inaccurate, biased, or out of date. Always maintain a "human-in-the-loop" approach by reviewing and fact-checking all outputs before use.
Professional Judgement: These suggestions do not substitute for formal legal, clinical, or safeguarding advice. Final responsibility for accuracy and appropriateness remains with the professional user.
