← Back to Blog
🎓 AI in Education

AI in Education Digest — August 2, 2026

Today's biggest story isn't a product launch — it's a policy, and it wasn't written by adults. While Congress has failed to pass a single national rule for AI in schools, 98 teenagers spent one weekend doing it themselves. Around that story: fresh data on how students actually use AI when they're struggling, a $100 million bet on AI tutoring from one of edtech's biggest names, and a compliance deadline that quietly moved for European universities today.

1. Congress hasn't passed AI rules for schools. So 98 teens did it themselves.

Source: NPR · July 30, 2026

Over one weekend in Boston, 98 high schoolers representing all 50 states gathered in a precise replica of the U.S. Senate chamber at the Edward M. Kennedy Institute and drafted the "Students First Act" — a model AI policy for K-12 schools. Hosted by the Kennedy Institute, AASA (The School Superintendents Association) and Day of AI, the teens split into subcommittees on students, educators, administrators and families before merging dozens of proposals into one bill.

The bill passed 82-16. It bans AI on graded tests, allows editing and brainstorming with AI after 8th grade, requires students suspected of misuse to defend their work through discussion or an oral exam, mandates that teachers post a written AI-use policy each semester, and — notably — requires school chatbots to alert a human adult, not just log the conversation, if a student appears to be in a mental health crisis.

Why it matters: With no federal AI-in-schools policy and only half of middle and high schools having any policy at all, AASA is sending this student-written bill to 10,000 school superintendents nationwide — meaning it could shape real classroom rules this school year.

The full text of the Students First Act is published alongside NPR's coverage — worth reading end to end.

2. Nearly three in five students turn to AI when sad, stressed, or lonely, JED research finds

Source: EdTech Innovation Hub, citing The Jed Foundation · July 30, 2026 (briefing held July 21)

Preliminary data from The Jed Foundation, previewed at a bipartisan Capitol Hill briefing, surveyed more than 5,500 middle and high schoolers across 21 U.S. schools. Four in five said they've used AI in some form, and nearly three in five said they've turned to it specifically while feeling sad, stressed, or lonely — more than one in ten used it for companionship.

Students who used AI for emotional support showed associations with poorer mental health, greater loneliness, and a greater history of suicidal thoughts — though JED is careful to frame these as associations rather than proof that AI use causes the outcomes, since the sample was drawn from schools already enrolled in its mental health programming. Full results are due this fall.

Why it matters: It's already shaping federal policy conversations about whether school chatbots should be required to route students in crisis to a real person — the same conclusion the teen-written Students First Act above arrived at independently.

The Jed Foundation's full study lands this fall — worth a follow-up video when it does.

3. Coursera invests $100 million in Andrew Ng's AI learning company LearnVector

Source: EdTech Innovation Hub · July 31, 2026

Coursera is putting $100 million into LearnVector, a new AI-native learning company founded by its own co-founder, Andrew Ng. Coursera CEO Greg Hart announced the deal on LinkedIn, describing LearnVector's ambition as building AI agents that give learners a "one-on-one learning experience" that adapts to how they learn and "stays with them until they've mastered the material."

The pitch is explicitly a break from both traditional one-to-many online courses and today's generative AI chatbots — instead of a search box that answers a question and moves on, LearnVector's agents are meant to keep working with a learner through practice and mastery. It builds on Coursera's AI push since its all-stock Udemy merger closed in May.

Why it matters: It's one of the largest single AI-tutoring investments of the year and signals where the "personalized learning" pitch is heading next — AI agents, not just smarter search boxes.

Ng has taught more people AI than almost anyone alive — worth watching closely what he builds next.

4. England's early-career teachers lead experienced peers in classroom AI use, Teacher Tapp data shows

Source: EdTech Innovation Hub, citing Teacher Tapp · July 31, 2026

New data from Teacher Tapp, covering the 2025-26 academic year, found England's early-career teachers (ECTs) use AI for more classroom tasks than experienced colleagues. In June, 68% of ECTs said they'd used AI for lesson planning in the previous month, compared with 54% of non-ECT classroom teachers — the largest gap in the dataset.

The gap held across other tasks too: ECTs were twice as likely to use AI for professional development (10% vs 5%) and three times as likely to use it for timetabling (6% vs 2%). ECT wellbeing also stayed consistently higher than experienced teachers' throughout the year, which Teacher Tapp links partly to the extra mentoring and reduced timetables new teachers receive — though it stops short of claiming causation.

Why it matters: It suggests the AI generational gap in teaching is real, and schools training new staff may need to rethink who's actually best placed to teach whom about AI tools.

Worth asking your newest teacher what they're using AI for this week — they may be ahead of the staffroom.

5. EU pushes back the AI Act's high-risk deadline for university assessment tools — from today to December 2027

Source: Travers Smith, DLA Piper GENIE, Responsible AI Labs · agreement finalized late June 2026

Today, August 2, 2026, was originally the date European universities needed to be fully compliant with the EU AI Act's "high-risk" rules for AI used in student assessment, admissions screening and academic progress monitoring — a category popular tools like ChatGPT don't currently meet. In May and June, EU lawmakers agreed to push that deadline back to December 2, 2027, as part of a wider AI Act simplification package.

The extra 16 months gives European universities meaningfully more runway to build in the transparency and human-oversight requirements the Act demands. Importantly, the deferral only moves the date enforcement starts — the underlying obligations themselves haven't changed, so institutions that assumed they were off the hook entirely should keep building toward compliance.

Why it matters: It's a genuine reprieve for a sector experts warned was nowhere near ready — but it's a delay, not a cancellation.

December 2027 will come around faster than universities think.

6. 90% of college students use AI in the classroom, Instructure poll finds

Source: Higher Ed Dive, citing Instructure · July 21, 2026

A new Instructure survey of more than 1,100 higher-ed students, instructors and parents found 90% of college students use AI in class at least occasionally, and 94% named at least one reason to be optimistic about it. Nearly two-thirds of instructors (61%) said the same about their own use.

But confidence in the tools hasn't kept pace with adoption: 65% of both students and instructors said they worry AI "can sound confident when it's wrong," and only 11% of instructors have received comprehensive AI training — 41% have had none at all. Every group surveyed said they expect schools to actively teach AI ethics and literacy.

Why it matters: Adoption has massively outrun training almost everywhere in higher ed — exactly the gap AI-literacy content and courses exist to close.

If 9 in 10 of your students are already using AI, the question isn't whether to teach with it — it's how.

🎬 Best story for today's video

Congress Couldn't Write AI Rules for Schools — So 98 Teenagers Did

This is the story with everything: a real narrative arc (Congress failed, teens stepped up), quotable soundbites straight from the Senate chamber floor, and a concrete artifact — an actual bill — that's now headed to 10,000 school superintendents. It's the most shareable, most human, and most "your kid's school policy could actually change" story of the week.

"While Congress still can't agree on a single word of AI regulation, 98 teenagers from all 50 states spent one weekend in a fake Senate chamber... and actually wrote the rules for how AI should work in your kid's classroom."

AI in education AI school policy ChatGPT in schools AI literacy Students First Act AI classroom rules EdTech news teacher AI tools AI regulation schools K-12 AI policy

Want tomorrow's digest before anyone else?

Get the daily AI-in-education briefing straight to your inbox — no fluff, just the stories that matter.

Subscribe to the newsletter →

🚀 Put today's AI in education to work

Revising for GCSEs or A-Levels? RevisionLab turns your notes into AI-powered practice questions, flashcards and personalised study plans — built for the way you actually learn.

As an Amazon Associate I earn from qualifying purchases.