← Back to Blog
🎓 AI in Education

AI in Education Digest — November 2, 2026

Dark navy graphic with the white serif headline Prove it's, and below it in teal, your own work, underlined in teal. On the right, an outlined exam paper card with grey text lines and a solid teal circle holding a navy tick. Footer: AI in Education Digest, 2 Nov 2026, derecktafuma.co.uk.

Seven stories, one thread: exams and schools are getting serious about proving the work is yours. Oxford’s investigations into AI cheating rose from one to 35 in three years, with capped resits and expulsions. South Korea will seek criminal charges for AI smart-glasses cheating before this month’s national exam, as a lab study finds AI can read blurred SAT papers. A London teacher and Micro:bit call for a GCSE in AI, as a survey finds only 15% of teens feel properly guided. A new AI tutor is trained to make students think, though it is so far tested only on simulated learners. And in the US, one district makes every assignment AI-free by default, Utah lets students help write the rules, and Wyoming offers a free AI-literacy course. The takeaway for families: do the first attempt yourself, and you never have to prove anything.

1. Oxford: AI-cheating investigations up 46% in a year

Source: Ned Remington, Cherwell (Oxford student newspaper) · 1 November 2026

A Freedom of Information request by Cherwell shows Oxford University ran 35 investigations into unauthorised AI use in assessments in 2025–26, up from 24 in 2024–25 and just one in 2023–24. Just over half of all such cases since September 2023 (31 of 60) came from the Social Sciences Division. Of the concluded cases, the most common outcome was a capped resit (28 times), where the best grade available is usually a pass; three students were permanently expelled. The rise comes after Oxford gave every student access to ChatGPT Edu last September.

Why it matters: AI cheating is no longer a hypothetical, and the penalty at a top university is real: a resit capped at a pass, or expulsion. The habit that avoids it starts at GCSE: do the first attempt yourself.

X reaction: No X posts about the Cherwell story were found (search for “Oxford AI plagiarism investigations” since 31 Oct returned 0).

Video angle: “One case three years ago. Thirty-five last year. Here’s what happens at Oxford if you’re caught using AI.”

Navy bar chart titled Oxford AI misconduct cases. Three bars: 2023 to 24, 1; 2024 to 25, 24; 2025 to 26, 35, the last bar solid teal. Source line: Cherwell FOI, investigations into unauthorised AI use in assessments, published 1 Nov 2026.

Oxford investigations into unauthorised AI use in assessments. Source: Cherwell FOI, 1 Nov 2026.

2. South Korea: criminal charges for AI smart-glasses cheating, as a study shows AI can read blurred exam papers

Source: Korea JoongAng Daily · 29 October 2026 72h+ window, labelled · study coverage: NoCut News / Yonhap · 1 November 2026 and The Asia Business Daily · 1 November 2026

Korea’s Ministry of Education says students caught using AI-powered smart glasses in the national university entrance exam (CSAT, on 19 November) may face police investigation and criminal charges, and it wants to extend the retake ban from one year to three. Every one of the 1,312 test sites will have a staff member dedicated to spotting smart glasses. It follows cases at Yonsei University and the University of Seoul, where applicants wore ordinary glasses at the pre-exam check and swapped to smart glasses during the test. On 1 November, Yonhap highlighted a German study (Urbanski and Peters, Martin Luther University Halle-Wittenberg) in which Google’s Gemini 1.5 Pro answered 82.78% of 720 past SAT questions correctly from deliberately blurred, angled images, and GPT-4o 75.69%. The researchers simulated the images; they did not test real glasses in a real exam hall.

Why it matters: Exam boards everywhere are watching. Expect tighter checks on glasses and devices, and more weight on work done under supervision. For students, the only safe edge is knowing it cold.

X reaction: Wide link-sharing, e.g. @channelstv (AFP) and @ennyola0015; @LinHindsight: “We used to worry about students hiding notes in their sleeves. Now the notes can be right in front of their eyes.”

Video angle: “They look like normal glasses. In a lab test, AI read blurred exam questions through them and got more than 80% right.”

Navy bar chart titled AI vs blurred SAT. Two teal bars: Gemini 83 percent, GPT-4o 76 percent. Small print: share of 720 past SAT questions answered correctly from distorted smart-glasses-style images, a lab simulation. Source: Urbanski and Peters, Martin Luther University Halle-Wittenberg, via Yonhap and NoCut News, 1 Nov 2026.

Lab simulation, not a real exam. Source: Urbanski & Peters via Yonhap/NoCut News, 1 Nov 2026.

3. UK: a London teacher, and Micro:bit, call for a “GCSE in AI”

Source: Metro · 1 November 2026

Metro profiles Kasia Suliga Wasielewska, a teacher at a north London school who builds AI into lessons: ChatGPT-generated quizzes, students fact-checking Gemini’s answers, and Micro:bit’s CreateAI tool to show young pupils how machine learning works. She says pupils as young as eight use AI for homework, and insists “AI must support, not replace, the teacher’s expertise.” The Micro:bit Educational Foundation wants the government to introduce an AI GCSE. Metro cites an Oxford University Press survey: more than nine in 10 teens use AI, but only 15% feel they’ve had enough guidance. The Department for Education told Metro it is co-designing AI tutoring tools with teachers and investing £23 million to find which technologies work best in schools.

Why it matters: Nearly every teen uses AI; very few have been taught how. Until schools catch up, the guidance gap lands on parents.

X reaction: @MetroUK shared it (4 replies at time of check); @OpenEyeComms quoted the Micro:bit call.

Video angle: “More than nine in 10 teens use AI. Only 15% say anyone has taught them how.”

4. humans& launches &Tutor, an AI model trained to teach

Source: humans&, “Introducing &Tutor” and model card · 30 October 2026 72h window

AI lab humans& released Ampersand Tutor (&Tutor) through its API. It was trained against “Persimmon”, a model that simulates a human learner, and rewarded by how much the simulated student learns. On TutorMoments, a benchmark built from real maths tutoring moments, it scored 0.754 under a plain prompt, ahead of Claude Sonnet 5.5 (0.667) and GPT-6 Astra (0.634); with an evaluation-aware prompt it scored 0.919, just behind GPT-6 Astra (0.926). The company’s own limitations section says results with simulated learners and AI judges “may not generalize to human learners” and that short evaluations “do not establish lasting learning gains.”

Why it matters: This is the right goal, an AI that makes you think rather than handing over answers. But the evidence so far comes from simulated students, as the company says itself. Treat it as promising, not proven.

X reaction: Launch post from @humansand; X’s trending summary (“New &Tutor AI Sparks Homeschooling Debate”) describes parents debating AI-led homeschooling, with critics worried about social skills.

Video angle: “A new AI was trained to teach by practising on a fake student. Here’s why that matters, and what it hasn’t proved yet.”

5. Howard County, Maryland: every assignment is AI-free unless the teacher says otherwise

Source: Kiersten Hacker, Baltimore Sun (via GovTech) · 30 October 2026 72h window

Howard County’s Board of Education approved an AI policy on Thursday 29 October, a week after Maryland’s deadline under its AI Ready Schools Act. All assignments are AI-free unless a teacher allows AI; when they do, students must follow set limits and disclose or cite their use. Staff may use AI for lesson planning, drafting messages or translation, but not for grading, counselling or entering personal data. Neighbouring Frederick County has banned student AI use in all grades, with limited exceptions. Student board member Mark Li said guidelines aren’t clear to students, and that when pupils can tell teachers used AI, they think, “Teachers used AI, so there’s nothing wrong if I do it.”

Why it matters: “AI-free unless told otherwise” is a simple rule families can copy at home for homework.

X reaction: @baltimoresun shared the story (3 quote posts at time of check).

Video angle: “One school district’s new rule fits on a fridge magnet: no AI unless your teacher says yes.”

6. Utah gives students a seat at the table on classroom AI rules

Source: Brooke Nally, KSL.com · 1 November 2026

The Utah State Board of Education has created an Artificial Intelligence Student Task Force, approved on 1 October, to feed student views into AI policy. It began with Skyline High School senior Rahul Nair, who read Utah’s AI framework, found little student voice in it, and wrote a charter for the task force. Hillcrest High senior Kennedy Mickelson joined with a more cautious view of AI. Nair’s question: if AI policies are being passed almost daily, why aren’t students being asked?

Why it matters: Rules work better when students help write them. That’s true in a district and at the kitchen table.

X reaction: No X reaction found (searched since 29 Oct).

Video angle: “Two students disagreed about AI in school, so they joined the group writing the rules.”

7. Wyoming offers every high school a free AI-literacy course and microcredential

Source: Wyoming Department of Education, superintendents’ memo for 2 November 2026 · page dated 28 October 2026 memo issue dated 2 Nov

Wyoming has partnered with Evergreen Education to offer Everwise, a self-paced AI literacy course of about 15 hours for grades 9–12, free to every high school. Students learn how AI tools work, where they go wrong, how to check what AI tells them, and when AI helps or gets in the way. The course neither pushes students to use AI nor to avoid it; each student writes a personal “AI Constitution.” Finishers earn a state-backed microcredential. An enrolment webinar is on 12 November.

Why it matters: This is the guidance gap from Story 3 being filled in practice: teach students to check AI, not just to use or avoid it.

X reaction: No X reaction found.

Video angle: “One US state is asking every teenager to write their own ‘AI constitution’.”

AI in educationAI cheatingexamsacademic integrityAI literacyAI tutorsGCSEparentsrevisionsmart glasses

Want this digest in your inbox? A short roundup of what’s actually changing in AI and learning.

Join the newsletter →

Do the first attempt yourself

Get The Revision Lab’s free GCSE History revision sheet: key people and dates for Black History Month on two A4 pages, with a timeline, exam tips and a quick quiz with answers, so your child can test themselves with no AI. The GCSE Maths Mock Booster is live.

Get the free revision sheet The Revision Lab