England's universities told AI has made coursework nearly impossible to grade fairly
A regulator-backed study of 127 institutions found AI use in essays and exams is so widespread that many assessments no longer reliably show what students actually know.
What happened: England's Office for Students, working with the charity Advance HE, interviewed more than 300 students and staff across 127 universities and colleges this year. The resulting report concluded that AI use has become so common in essays, dissertations and take-home exams that staff described grading some coursework as close to impossible. The regulator warned institutions that failing to build assessments AI can't easily fake could breach their conditions of registration, opening the door to sanctions.
Why it matters: Degrees exist to certify that a graduate actually has certain knowledge and skills. If assessments can no longer reliably tell who did the work, that certification starts to lose meaning for employers, other universities, and honest students sitting next to people who didn't do the reading. The OfS called this a generational risk to the whole system, not a one-off cheating scandal.
How it works, plainly: Students told researchers that classmates routinely finish assignments with ChatGPT in minutes, submit them, and score similarly to peers who wrote unaided. Some couldn't recall basic facts about work they'd just handed in. Staff separately noticed batches of essays citing identical sources in near-identical phrasing. Because the cheating usually isn't crude copy-paste, traditional plagiarism checks often miss it entirely, and once one attempt goes undetected, students told researchers they simply repeat it on the next assignment.
The rollout: Many universities are already shifting weight back onto formats AI can't easily do for you: in-person written exams, oral vivas where students must explain their work live, and process-based assessment that tracks how a piece of work was actually researched and drafted. Universities UK, representing 147 institutions, says substantial changes are underway but acknowledges there's further to go. The Department for Education says it expects the OfS to make its AI rules explicit and enforce them, while the regulator says it will keep working with the sector rather than issuing blanket bans.