AI in Education Digest — August 22, 2026

The loudest story in AI and education this week isn't a product launch — it's a demand that universities scrap an entire category of exam. Four-fifths of UK institutions still set assessments students sit at home, unsupervised, with a frontier model one tab away. A report published this week says end that immediately. Around it, the proctoring industry is arriving with the fix, OpenAI is attaching student credits to a data centre deal, and three separate programmes are trying to teach people about AI before it quietly teaches them. Seven stories that matter.

1. UK universities told to ban unsupervised remote exams — immediately

Source: Times Higher Education, reporting Policy Exchange · 18 August 2026

Thinktank Policy Exchange has told UK universities to stop using unsupervised remote exams straight away, arguing that continuing risks undermining public confidence in the value of a degree. Freedom of Information requests sent to 120 UK universities found that 78 per cent rely on online, remote exams for summative assessment. Only 10 per cent of institutions use online invigilation for all such tests, and around a third of university policies covering these exams do not mention generative AI at all. Despite that, 70 per cent say they plan to carry on. More than 90 per cent of students report being assessed by take-home essays, and 55 per cent by remote online exams.

The report — Evidence of Learning Through Assessment: Protecting the Value of a Degree in the Age of AI, authored by Swansea University neuroscientist Phil Newton — argues that existing problems including grade inflation, essay mills and inadequate staff training have been "tipped into a full-blown crisis" by the ubiquity of AI. Its sharpest point is about fairness rather than technology: students who follow the rules are punished with lower grades, then graduate into a competitive job market carrying the same debt as classmates who did not. Neil Mendoza, provost of Oriel College, Oxford, writes in the foreword that universities "must act now to put their house in order, before it is too late." The recommendations go beyond banning unsupervised tests — a return to in-person assessment, more detailed degree classifications showing what a graduate actually learned, and independent evidence-based training standards for university educators. Former Liberal Democrat leader Vince Cable also endorsed it, calling for in-person examination to replace remote assessment "unless the latter is fully secure."

Video angle: "78% of UK Unis Still Use Online Exams. A New Report Says Ban Them Today."

2. Synap launches AI-assisted proctoring one day after the report

Source: EdTech Innovation Hub · 19 August 2026

Leeds-based assessment company Synap has released Synoptic, an integrated proctoring system for education providers, training organisations and professional bodies. It combines exam delivery and invigilation in one platform: sessions are recorded, an audit trail is created, and AI highlights potentially suspicious activity — leaving full-screen mode, opening a new tab, moving focus away from the exam window — for a human invigilator to review. Reviewers can move between an exam reconstruction, webcam images, screenshots and a timeline of events. Monitoring is configurable by risk rather than applied uniformly, with options including live supervision, post-exam review, periodic screenshots and HTML reconstruction for low-bandwidth settings.

The timing is not accidental: the UK launch was deliberately placed between A-Level and GCSE results days, as regulators weigh how generative AI and wearable technology affect assessment integrity. Synap cites Ofqual Chief Regulator Ian Bauckham's comments that detection rates for AI in assessed work remain low despite teacher concern, and notes the Association of Chartered Certified Accountants has withdrawn remote invigilated exams over cheating fears. Crucially, the company does not claim AI makes the final call. "The conversation shouldn't be whether online exams can be trusted," said CEO Dr James Gupta. "It should be whether institutions are using technology that deserves to be trusted." The platform is encrypted and Synap says it adheres to ISO/IEC 27001 standards.

Video angle: "The Exam Software That Watches You Change Tabs — And Who Decides If You Cheated"

3. OpenAI ties an 8-gigawatt Ohio data centre to $84m of student Codex credits

Source: EdTech Innovation Hub · 19 August 2026

OpenAI has agreed to secure roughly eight gigawatts of computing capacity at the planned PORTS-Pike Technology Data Center in Pike County, Ohio, on a 20-year lease from developer SB Energy, hosting NVIDIA infrastructure exclusively. The first 800 megawatts is expected to be available in 2028. Bolted onto the deal is an education offer: up to $84 million in Codex credits for approximately 844,000 eligible Ohio students. College, community college and technical school students aged 18 and over get $100 through their ChatGPT accounts during the 2026–2027 academic year. OpenAI has also committed $40 million to a locally shaped grant fund, adding to $40 million already pledged by SB Energy — more than $160 million in proposed benefits in total, by the company's own framing.

The jobs numbers are large: a six-year buildout through 2032 expected to create 35,000 construction jobs, then 2,500 long-term operating positions, with a Memorandum of Understanding already signed with North America's Building Trades Unions. Property tax revenue is expected to flow mainly to public schools, though no precise figure has been given beyond "hundreds of millions" over the project's lifetime. Chris Lehane, OpenAI's Chief Global Affairs Officer, was unusually direct about the scepticism: "Projects at this scale raise legitimate questions about electricity costs, water use, jobs, and whether the promised benefits will actually reach local communities." OpenAI says it will publish annual figures on local recruitment, community investment, water consumption and energy use. Expected water use has not yet been published because the site design is unfinished, and growth beyond 800 MW requires new grid-connected generation including natural gas.

Video angle: "OpenAI Is Giving 844,000 Students $100 Each. Here's What It's Actually Buying."

4. Crisis Text Line joins Common Sense Media's Youth AI Safety Institute

Source: EdTech Innovation Hub · 20 August 2026

Crisis Text Line has joined Common Sense Media's Youth AI Safety Institute as a founding standards partner, bringing evidence from real crisis conversations into the safety criteria used to assess AI products aimed at young people. The partnership focuses on three severe risks: self-harm, suicidality and emotional manipulation. The organisation has trained more than 55,000 volunteers since launching in 2013 and provides free, confidential text-based crisis support in English and Spanish, 24 hours a day.

The specific contribution is about how distress actually shows up. Common Sense Media says Crisis Text Line's experience will inform how products are evaluated when a young person signals difficulty indirectly, ambiguously, or gradually across a long conversation — rather than only through a narrow list of high-risk keywords. "Crisis Text Line has learned from millions of crisis conversations how young people signal distress," said Robbie Torney, Head of AI and Digital Assessments at the Institute. "That experience will help us measure AI products against how kids really behave when they're at risk." One honest gap remains: the Institute argues AI must respond safely to less direct signals but has not yet specified what an acceptable response looks like. Human support stays separate from the AI — Crisis Text Line connects people to live trained volunteer counsellors, with supervisors holding relevant master's degrees or comparable crisis intervention experience.

Video angle: "Why AI Safety Filters Miss the Teenagers Who Need Them Most"

5. CodeAI and OpenAI partner on AI literacy and a student Builders Challenge

Source: EdTech Innovation Hub · 20 August 2026

CodeAI and OpenAI have formed a partnership running across several programmes over the next year. OpenAI will support CodeAI's Hour of AI, contribute expertise to the year-long AI Foundations high school course, and provide mentors for CodeAI's first Builders Challenge. A joint advisory council will bring together specialists in child development, youth public policy and learning science to advise on emerging risks and responsible AI practice, though its membership has not yet been announced. A separate Career Journeys strand will introduce students to OpenAI researchers and engineers.

The numbers behind it explain the urgency. CodeAI says only 16 per cent of high school leaders believe all of their students are learning the technical knowledge needed to understand AI in the classroom, while its own research found 75 per cent of high school students think understanding AI will matter more in future than it does today. CEO Karim Meghji framed the goal as a "glass box" rather than a black one: "Every student should know how AI actually works and be able to question the technology, catch its mistakes, and know when to stop trusting it." Dr Allison Fine Mishkin, Head of Child Development at OpenAI, said the aim is "not simply to help more students use AI, but to help them ask better questions." Details that are still missing: eligibility, dates and applications for the Builders Challenge, and exactly what OpenAI's input will change in AI Foundations.

Video angle: "The One AI Skill Schools Aren't Teaching (And Why No Company Will Teach It)"

6. UK government starts an AI summer bootcamp for North West school leavers

Source: EdTech Innovation Hub · 18 August 2026

Young people from Wigan, Blackburn, Blackpool and other North West communities have begun AI training through the UK government's new TechFirst AI Apprenticeship Pathway, described by government as the first AI summer bootcamp of its kind and aimed at reducing post-school unemployment. Delivery is by IN4 Group, bringing together government, businesses and public sector organisations across the region. JD Sports and Heinz are among the employers offering apprenticeships through the pathway, and Secretary of State for Culture, Media and Sport Lisa Nandy met recruits taking part. "AI is already shaping where jobs go," wrote Kanishka Narayan MP, the UK's Minister for AI. "The best preparation for our young people will be the skills to thrive with AI."

The gaps are worth naming. The announcement does not state cohort size, bootcamp duration, curriculum content or selection criteria, nor how many apprenticeships each employer is offering — and there is no confirmation that everyone completing the training moves into an apprenticeship. No application route or timetable for future cohorts has been published either. IN4 Group itself is well credentialed: approved training provider status with Microsoft, Amazon Web Services and Google, rated "Good" by Ofsted, and company figures citing work with more than 500 employers, over 2,000 schools and colleges, and more than 6,000 people placed into technology jobs. On 12 August it also announced a partnership with the Electech Innovation Cluster to train electronics employees across Lancashire and Cumbria via the Level 4 AI and Automation Practitioner apprenticeship standard.

Video angle: "The UK Just Started AI Bootcamps for School Leavers — What They Don't Tell You"

7. Frontier Fellowship opens 250 free AI training places for women

Source: EdTech Innovation Hub · 20 August 2026

Applications have opened for the Frontier Fellowship, a 12-week programme offering 250 women in England free training and live business experience to prepare them as AI Deployment Consultants. It is delivered by Girls Talk Corporation and Enterprise4all, funded by Innovate UK under the Department for Science, Innovation and Technology through the TechLocal programme. Two cohorts start in October and November 2026, with applications open during August. Eligibility is women aged 18 and over living in London or the North West who are unemployed, returning to work, or employed with insufficient hours or limited progression, and who are not in full-time education. No degree, technology background or prior AI experience is required — "We are not looking for experience. We are looking for potential."

The structure is unusually concrete. Weeks 1–4 cover tools including Claude, Microsoft Copilot and ChatGPT plus workflow building and business process mapping. Weeks 5–8 put teams of four or five on a four-week project for a UK SME with a named client, defined scope and weekly milestones, drawn from eight planned project types across sales, finance, operations, HR and marketing. Host businesses pay nothing but must supply a named project lead for at least four hours a week plus system access. Weeks 9–12 focus on employment: interview coaching, CV preparation, employer introductions and Insight Days, followed by six months of alumni support. Selection involves no CV — an online form, a roughly ten-minute scenario assessment, then a 45-minute case study for shortlisted candidates. The honest caveat: the programme guarantees training and a live project, not permanent employment, though Girls Talk cites an independently assessed 80 per cent sustained job outcome rate from previous programmes.

Video angle: "A Free 12-Week Route Into AI Work — No Degree, No CV, 250 Places"