If you blinked this week, you missed a lot. OpenAI just paid 26 students to build the future, Google quietly turned NotebookLM into a research agent, a $8 million tutoring fund landed with the Gates Foundation behind it, and Anthropic made its AI courses free for anyone with a learner's mindset. Here are the six stories actually worth your attention today — and why each one matters if you're a student, a teacher, or a parent trying to keep up.
OpenAI hands 26 students $10K each to build the future of AI
OpenAI named its inaugural ChatGPT Futures Class of 2026 — 26 young people and teams, each receiving a $10,000 grant plus access to OpenAI's frontier models. This is the first university cohort to have had ChatGPT available for their entire degree, and what they're building is a long way from "write my essay."
The standouts are heavily education-flavoured: Crystal Yang, 18, is building audio-first learning games for 200,000 blind and visually-impaired students; Shraman Kar, 19, makes learning videos in dozens of languages; and a duo is scaling an AI college counsellor to students in 190 countries. Others are matching students to scholarships and turning CS explainers into a 12,000-member nonprofit.
The future isn't coming for these kids. They're the ones shipping it.
Google rebuilds NotebookLM into a research agent with Gemini 3.5
Google upgraded NotebookLM — the study tool that turns your own sources into summaries, study guides and audio overviews — with Gemini 3.5 and a set of new agentic research features. Translation: the app that already made revision notes from your PDFs can now go and do multi-step research legwork on its own.
For anyone who's used NotebookLM to cram for an exam or untangle a dense reading list, this is a meaningful jump. Smarter reasoning plus agentic tools means tighter summaries, better-connected ideas, and a study partner that can chase down the gaps instead of waiting for you to spoon-feed it every source.
The free study tool just got a postgraduate upgrade.
A $8M AI tutoring grant opens — with the Gates Foundation behind it
Digital Promise opened an $8 million grant programme, backed by the Gates Foundation, to fund AI tutoring projects. It's a signal that serious money now sees AI tutoring as core infrastructure for schools rather than a side experiment — and that funders want evidence it actually moves learning outcomes.
The timing fits a broader shift: districts are formalising AI tutoring into before-school, after-school and at-home support, trying to give every learner the kind of one-to-one help that used to be a luxury. The open question is whether grant-funded pilots can prove real gains, not just engagement dashboards.
Free, personal tutoring for every kid has always been the dream. This is someone funding the test.
Anthropic makes its AI fluency courses free on Coursera
Anthropic brought its AI fluency courses to Coursera, free for students, educators and nonprofits. As AI becomes part of everyday coursework and admin, "how do I actually use this well?" has become the real skills gap — and the company behind Claude is betting that teaching fluency, not just access, is where the value sits.
For teachers especially, free, structured training is a big deal. Most AI professional development is either expensive, vague, or both. A no-cost course aimed at the people who set classroom norms could do more for responsible AI use than any policy memo.
Access was never the hard part. Knowing what to do with it is.
University of Leicester rolls out Microsoft 365 Copilot across campus
The University of Leicester is rolling out Microsoft 365 Copilot across the whole institution, putting an AI assistant inside the Word, Outlook and Teams that students and staff already use every day. It's one of the clearer examples of a UK university treating AI as standard kit rather than a banned distraction.
Campus-wide deployments like this matter because they normalise AI as part of the workflow, with institutional guardrails, rather than something students sneak around with. The next question every university faces: how do you assess learning when the assistant is built into the tools?
The debate about whether AI belongs on campus is quietly being decided by IT rollouts, not op-eds.
AI in the classroom hits the statehouse: 134 bills, 31 states
While the tools race ahead, lawmakers are scrambling to catch up: 134 AI-in-education bills have been introduced across 31 states this session, focused on data privacy, classroom-use limits and curriculum. California's AB 1159 would ban using student data to train AI models, Idaho's SB 1227 mandates data-privacy protections, and states like Oklahoma and Maryland want human oversight on any high-stakes decision about a student.
It's the messy, necessary other half of the AI-in-education story. For every flashy product launch, there's a parent asking who sees their kid's data and whether a model is quietly grading or sorting their child. Expect "human in the loop" to become the phrase of the school year.
Innovation makes the headlines. Regulation decides what actually reaches the classroom.
OpenAI Just Gave 26 Students $10K to Build the Future of AI
It's aspirational, concrete and endlessly clippable — real young people, real grants, real tools helping blind students and first-gen scholarship seekers. It flips the tired "AI is for cheating" narrative on its head, which is exactly the kind of angle an education audience shares.
Get this digest in your inbox 📬
Five minutes a day on how AI is really changing education — no hype, no overwhelm.
Join the newsletter →