← Back to Blog
🎓 AI in Education

AI in Education Digest — August 24, 2026

The back-to-school AI arms race has gone from marketing to product. In the past fortnight Google has rebuilt Search and Gemini around studying, Anthropic has opened a free public academy, four more US states have made district AI policies compulsory, and a room full of teenagers has written a national AI rulebook that is stricter than most schools would dare propose. Here are the seven stories worth your attention.

1. Google turns Search and Gemini into a study tutor

Source: TechCrunch (Aisha Malik) · 19 August 2026

Google has shipped a wide set of study features across Search and Gemini. In Search, students can now generate interactive visuals and simulations inside AI Overviews — search "pH scale" and get a manipulable diagram rather than a paragraph — and ask follow-ups that spin up a tailored interactive experience. Search will also generate custom practice quizzes on any subject, and can build study documents from uploaded PDFs, slides or a photo of handwritten notes.

In Gemini, students can kick off multi-step research reports that run in the background and then discuss the findings by voice in Gemini Live. Gemini can also render functional 3D simulations — rotate and zoom a DNA helix, for example — and a new dedicated hub in the app pulls study notebooks, flashcards and practice quizzes into one place. A Lens-based feature that photographs a problem and walks you through it is rolling out in the coming weeks.

Why it matters: This moves AI from "write it for me" to "explain it to me" — the first genuinely defensible use case for AI in revision, and the one teachers have been asking for.

Video angle: I replaced my revision notes with Google Search for a week.

2. Anthropic opens Claude Academy to everyone, free

Source: EdTech Innovation Hub · 20 August 2026

Anthropic has opened Claude Academy, a free public learning hub at academy.claude.com built around the way the company trains its own staff. Reporting on the launch describes a problem-first curriculum rather than a feature tour, organised around a "4D AI fluency" framework, spanning a first conversation with a chatbot all the way to building on the Claude API, Claude Code and agents.

Coverage puts the catalogue at roughly two dozen courses alongside a large library of tutorials, documented use cases and webinars. The material is free and readable without signing in; users with a Claude account can track progress and earn completion badges they can add to LinkedIn. It follows Claude for Teachers, which launched free for verified US educators in July.

Why it matters: AI literacy is quietly becoming a credential. When the model labs give away structured training, "I use AI" stops being a differentiator and "I can prove it" starts being one.

Video angle: Anthropic just made its internal AI training free — here is what is actually in it.

3. The education AI gold rush comes with a bill attached

Source: The Christian Science Monitor (Aaron Glantz) · 7 August 2026

A Monitor investigation puts hard numbers on the classroom AI boom. Figures prepared for the paper by Future Market Insights size the education AI sector at 730 million dollars in 2026, growing around 40 percent a year to a projected 18.5 billion dollars by 2036. The Center for Democracy and Technology estimates 85 percent of teachers and 86 percent of students already use generative AI for schoolwork, while more than 30 states have written their own guidance in the absence of federal standards.

The reporting is blunt about what is not established. Chris Agnew of Stanford's Generative AI for Education Hub tells the Monitor that the efficacy of personalised-learning chatbots has not been independently demonstrated, and that his centre documented a performance drop once the AI was taken away. Common Sense Media rated the schools version of Google's Gemini "high risk". A February University of Pennsylvania Wharton study of 1,300 participants across 10,000 trials described AI use as reshaping human reasoning and producing what the authors call cognitive surrender. The piece also notes that the free educator tiers have expiry dates — Anthropic's in June 2027, OpenAI's in June 2028.

Why it matters: Free tools with expiry dates are a customer-acquisition strategy, not a gift. Schools building workflows on them today are negotiating from a weak position in two years.

Video angle: The AI tools your school gets free right now have an expiry date.

4. Students from all 50 states write their own AI law — and it is strict

Source: AASA, The School Superintendents Association · 3 August 2026

At America's Youth AI Festival, held 17–19 July on the UMass Boston and MIT campuses, students from every state formed a National AI Student Senate in the Kennedy Institute's replica Senate chamber and passed the STUDENTS FIRST Act of 2026 by 82 votes to 16. It rejects both an outright ban and unrestricted adoption.

The framework would prohibit AI-generated written or artistic assignments at every grade level, permit AI from ninth grade for brainstorming, studying and editing with teacher permission and disclosure, and bar it entirely on graded assessments. It also guarantees students a right to appeal an improper-use accusation with human review, stops schools using AI alone to decide grades, discipline, profiling or hiring, and bans AI deepfakes used for harassment. The event was organised by Day of AI, MIT RAISE, AASA and the Kennedy Institute, and AASA will circulate the framework to more than 10,000 school leaders.

Why it matters: Students have written a tighter policy than most adults would have dared, and they built in due process. That combination is hard for any district to argue with.

Video angle: Students wrote their own AI rules — and they are stricter than your school's.

5. Do law school AI bans teach the wrong lesson?

Source: University of Maryland, Baltimore / Jurist commentary · 6 August 2026

Berkeley Law has introduced a policy that, unless an individual professor opts out, bans almost all AI use in coursework and exams. Students may not use AI to conceptualise, outline, draft, revise, translate or edit anything submitted for credit — and may not even ask a tool to fix a grammatical error. Narrow source identification is the exception.

Writing in Jurist and republished by UMB, Maryland Carey Law professor and former dean Donald B. Tobin and JD student Samuel Irwin argue the policy exports the learning curve to employers. They note that courts have now addressed AI-fabricated filings in well over a thousand decisions, including a six-figure sanction this spring, and argue that graduates who have never been taught to supervise AI output are exactly the people who generate those filings. They favour assignment-level rules over blanket institutional bans, and single out the grammar and spell-check prohibition as especially harmful to students with dyslexia. The authors disclose that AI tools assisted in drafting the essay.

Why it matters: The debate has moved past "should students use AI" to "who is responsible for teaching them to use it safely" — and abstinence policies quietly answer: someone else.

Video angle: The AI ban that could make lawyers worse at their jobs.

6. A 45,000-person survey finds universities losing the AI race to their own students

Source: EdTech Innovation Hub, reporting the Digital Education Council Global Survey 2026 · 9 July 2026

The Digital Education Council's AI in Higher Education Global Survey 2026 draws on 45,398 responses — 27,284 students and 18,114 faculty across 35 countries. It reports that 88 percent of students now use AI in their learning and 77 percent of faculty use it in teaching, the latter up 16 percentage points on 2025.

Governance is the gap. Fifty-seven percent of students say assessments arrive with inadequate AI guidance, only 29 percent believe their instructors are equipped to guide AI use, and just 31 percent of faculty say their institution involves them meaningfully in AI policy. The dependence signals are notable too: 22 percent of students find it harder to work without AI, 19 percent say they retain less, and 66 percent globally — 81 percent in the US and Canada — worry AI is making learning too shallow. Strikingly, 43 percent of US and Canadian students say they would support an institution-wide AI ban.

Why it matters: Adoption is effectively universal; policy is not. The students asking for a ban are not technophobes — they are the heaviest users, asking for guardrails nobody has given them.

Video angle: 43 percent of students want AI banned — and they are the ones using it most.

7. Four more US states now require districts to adopt AI policies

Source: K-12 Dive (Anna Merod) · 9 July 2026

New laws in Idaho, Maryland, Oklahoma and Virginia require state education departments to publish guidance on safe, responsible AI use and oblige school boards to adopt aligned policies of their own. Ohio and Tennessee passed similar mandates earlier.

The detail differs by state. Oklahoma bars AI tools from being used primarily for grading, discipline or other high-stakes decisions, lets parents opt their children out without academic penalty, and requires districts to disclose annually which AI tools they use and what student data is collected. Maryland requires every district to designate an AI coordinator, tasks the state with building an evaluation rubric and professional development, and creates a statewide AI Education Collaborative. Virginia adds a pilot programme. For context, K-12 Dive notes that as of October 2025 thirty-four states plus Puerto Rico had issued AI guidance, while a May Gallup and Walton Family Foundation survey found 82 percent of teachers had received no formal AI guidance at all.

Why it matters: Guidance is becoming statute. The interesting bit is the shift from "how should we use AI" to "what may AI never decide" — grades, discipline, and who gets hired.

Video angle: These four states just made AI policy the law in every school.

🎬 Best story for today's video

Google turns Search and Gemini into a study tutor

It is the highest-intent story of the week: every student watching has a Google search bar open right now, the features are live today, and it is demonstrable on screen in under sixty seconds. The other stories are policy — this one changes what a viewer does tonight.

Google just quietly turned the search bar into a tutor. Interactive 3D models, instant practice quizzes, and a study hub that will read a photo of your handwritten notes. If you are still revising the way you did last year, you are already behind — but there is one catch nobody is mentioning.

Three-point outline for the video

1. What actually shipped. Interactive visuals and simulations inside AI Overviews, custom practice quizzes on any subject, Lens photo-to-explanation for when you are stuck, study documents built from your own PDFs and handwritten notes, and a dedicated hub in the Gemini app with notebooks and flashcards.

2. Why Google is doing this now. This is a distribution war, not a charity drive. Google, OpenAI and Anthropic are all racing to be the assistant a student opens first, and free student tiers are the weapon — with expiry dates attached, as the Monitor reported this month.

3. The catch. The research keeps saying the same thing: faster completion is not the same as more learning. Use the tools to test what you know, not to skip knowing it. Practical rule — generate the quiz, then close the tab before you answer it.

AI in education Google Gemini AI study tools Claude Academy EdTech AI school policy GCSE revision A-Level study AI tutor learning with AI

Want this in your inbox every morning?
One short email, seven stories, no hype — the AI in education brief for students, teachers and parents.

Join the newsletter

🤖 Got the ideas but nowhere to put them?

I build fast, clean websites for tutors, freelancers and small businesses — the kind that turn your thinking into enquiries.

Book a Consultation →

Building your reading stack? Browse top-rated books on learning and note-taking on Amazon. As an Amazon Associate I earn from qualifying purchases.