Ofqual Won't Rule Out Scrapping Coursework Over AI Cheating

A regulator runs out of patience

Sir Ian Bauckham, Ofqual's chief regulator, has said something schools should take seriously: getting rid of coursework altogether is "never off the table." Speaking to Schools Week in April, and again in Ofqual's new podcast Can I Just Qualify That? in June, Bauckham has been unusually blunt about the state of non-examined assessment (NEA) in the age of generative AI.

This is not idle musing. In March, Bauckham wrote to the chief executives of England's exam boards demanding tougher action on AI misuse in coursework. He is now evaluating their responses, and by his own account, one option on the table is simply removing extended, unsupervised writing tasks from qualifications where they still exist — including the extended essays that make up around a fifth of some history and English A-level grades.

"If you genuinely believe that the process of researching and completing an assignment is an important part of the learning process," Bauckham has said, alternatives such as checkpoints and stricter source referencing are worth trying — but abolition, he added, "can never be off the table."

Why this is different from a detection problem

Much of the public conversation about AI and coursework has focused on catching cheats: similarity scores, watermarking, JCQ guidance on what counts as malpractice. Bauckham's position is more structural. He has told MPs plainly that AI marking tools are not reliable enough to be trusted either, and the deeper issue is that AI-generated writing is becoming close to indistinguishable from a student's own work when produced at home, unsupervised, over days or weeks.

Ofqual's own malpractice data gives the scale of the wider problem. In the summer 2025 exam series, 2,225 cases of malpractice involved mobile phones or smart devices — 44.3 percent of all student misconduct, and the largest single category every year since 2018. Bauckham has separately flagged smart glasses and invisible earpieces, now openly advertised on social media, as a fast-growing threat even inside exam halls. Coursework, done at home with no invigilation at all, is the much softer target.

That is why "getting rid of NEA" is being discussed not as a punishment for cheats but as an admission that authenticity can no longer be guaranteed under the current model.

What actually gets cut

England already has comparatively little coursework left in its qualifications, after successive reforms since 2015 stripped it out of most GCSEs. What remains tends to be concentrated in subjects where a written, research-based task is hard to replace with a written exam: extended essays in history and English literature A-levels, non-examined assessment in art and design, drama, music and some vocational and technical qualifications. These are precisely the subjects now under Ofqual's microscope.

If NEA disappears from these courses, the likely replacement is not "an exam instead" so much as a redesigned assessment: shorter, supervised research tasks with checkpoints; in-class writing windows; oral defence of a piece of work (a viva-style check); or simply moving marks back into a timed, invigilated paper. None of these are neutral choices. They change what gets taught, how much classroom time assessment consumes, and which pupils are advantaged — pupils who write fluently under exam pressure are not always the pupils who produce the best researched, redrafted coursework.

The pressure is coming from more than one direction

This sits alongside a broader adoption gap documented in the Digital Education Council's 2026 global survey of nearly 45,400 students and faculty across 35 countries, which found AI use among students has reached 88 percent, up sharply on last year, while only 29 percent of students believe their instructors are properly equipped to guide them on using it responsibly. Coursework authenticity is the sharpest edge of a much wider mismatch between how fast students are adopting AI tools and how slowly assessment systems are adapting.

Ofqual's dilemma, in other words, is not unique to England. But because Ofqual regulates a system that still contains meaningful chunks of unsupervised assessment, it is one of the first regulators to face the choice explicitly and in public.

What to do

School leaders should not wait for Ofqual's final decision before acting. Exams officers and heads of department in affected subjects should already be identifying which qualifications include NEA, extended writing or on-screen coursework, and building in checkpoints — drafts, source logs, supervised writing sessions — that create a paper trail of a pupil's own process rather than relying on a finished product alone. Departments that treat coursework as something pupils "take away" rather than something staff actively witness being built are the most exposed if reform arrives mid-cycle.

What to watch

Watch for Ofqual's response to the exam boards' submissions, expected later this year, and for any subject-specific consultations on removing or redesigning NEA. Also watch whether other UK regulators and awarding bodies — and international counterparts grappling with the same Digital Education Council findings — start floating similar structural changes rather than incremental detection fixes. If England moves first, it will be a template other systems copy or reject.

Sources: