Call to force AI firms to help universities catch essay cheats

New tools designed to help students rewrite artificial intelligence-generated essays should prompt a radical rethink in regulation, with platforms required to work with universities on tackling plagiarism, a legal expert has urged.

At the start of the new academic year, universities have been warned about an explosion of companies offering low-cost services to evade cheating checks, many of which are being aggressively marketed to “anxious students” on TikTok, Facebook and Instagram, said Michael Veale, associate professor in digital rights and regulation at UCL.

Some companies are boasting blogs with titles such as “How to bypass Turnitin” and “How to write an AI-resistant essay”, research by Dr Veale and colleague Noëlle Gaumann has found.

This was a sign that essay mills have pivoted from offering made-to-order essays to providing tools to disguise AI-written content, Dr Veale told Times Higher Education.

In some cases, “institutional plagiarism checkers seem to be playing both sides of the market”, with some large edtech firms providing both a “premium AI…to rephrase AI generated or normally plagiarised work so that it can avoid detection” and a plagiarism detector.

Legislation in England aimed at eradicating contract cheating should be updated to force AI firms to work with university authorities, said Dr Veale.

Campus resource: Can we spot AI-written content?

“Educational providers could be offered, within examination periods, the ability to pass examination or problem questions securely to an [AI services] provider,” he said, with firms assessing the likelihood of plagiarism using their tools.

Technology providers should “not seek to ban or block such queries, as this would be in the direction of internet switch-offs for exams…but should instead retain the results of these queries in an answer bank which licensed plagiarism detection tools have access to as part of the corpus”, said Dr Veale.

He added: “More importantly, universities need to stop being fatalist, flaccid rule-takers around technologies – the current leadership seem to feel they have no ability to drag these companies to the table and obtain concessions and governance mechanisms from them. This needs to change.”

Academics lamented being swamped by mediocre AI-written essays during this summer’s marking season, with many unconvinced by a shift away from AI bans towards asking students to declare AI use.

“There are the statements of generative AI use that some universities are requiring on assessments, and also more specific guidance on assessment briefs, but it’s still rather variable,” commented Thomas Lancaster, an academic integrity expert based at Imperial College London.

“I’m still seeing questionable practices, like requiring students to quote and reference GenAI text and to provide copies of the chats, which is unworkable in many situations, as a student may be using multiple chats and different systems.”

jack.grove@timeshighereducation.com