UK universities rethink exams as ChatGPT sparks cheating fears

The university essay might never look the same again.

Just weeks after OpenAI released ChatGPT to the public, lecturers across UK universities are sounding the alarm over assessments. With the chatbot capable of generating high-grade academic essays from a simple prompt, academic staff are now forming emergency working groups to reconsider how coursework is graded.

Banning essays and rethinking exams

Some departments have already stopped waiting for guidelines and made their move.

At University College London, staff in the computer science department decided to eliminate the essay option entirely from their final coursework. Students previously had a choice between an essay-based or skills-based assessment; now, only the skills track remains.

Across the globe, institutions are scrambling in different directions. In the US, New York City public schools outright banned ChatGPT on all school devices and networks. Australian universities have signaled a potential retreat to old-fashioned, in-person written exams to stop unearned grades in their tracks.

Dr. Thomas Lancaster, an academic integrity researcher at Imperial College London, called ChatGPT a major turning point. The underlying tech isn't brand new, he noted, but wrapping it in an effortless chat interface makes it accessible to anyone overnight. Still, Lancaster cautioned against a total regression into exam halls, advocating instead for oral exams and live presentations.

There is also the plain fact that the bot still hallucinates with confidence. Michael Draper, an expert on legal education and academic integrity at Swansea University, warned that the more specific and fact-heavy an essay prompt is, the more likely ChatGPT is to produce polished nonsense. Even so, Draper argued that an outright ban misses the point if workplaces are already adopting these tools.

Michael Webb, director of technology at Jisc, echoed that sentiment, comparing ChatGPT to a next-generation spellchecker. Rushing to block it is neither realistic nor productive; instead, schools should rethink assessments to judge critical reasoning over raw essay assembly.

Why it matters

For students and educators, the ground has shifted under traditional evaluation. Attempting to outsource homework to ChatGPT might work in the short term, but relying on it is playing with fire given its tendency to fabricate plausible-sounding facts. More importantly, students will soon face defense vivas and skills-based hurdles where AI-generated fluff offers zero cover.

For academia at large, the traditional take-home essay is fundamentally broken as a measure of understanding. Plagiarism checkers rely on matching existing strings of text, meaning uniquely generated AI output slips right through the gate. This forces an overdue overhaul of how schools measure actual intelligence versus mere information regurgitation.

In our view, retreating to paper exams or slapping network-level bans on school Wi-Fi is short-sighted security theater. You cannot put this genie back in the bottle, especially when graduates will be expected to use AI on the job. The smarter pivot is acknowledging the tool exists and redesigning prompts around debate, source interrogation, and oral defenses that test whether a student actually understands the words on the page.