TL;DR

Suspected or confirmed cheating in graduate recruitment has reached a record, with 92% of employers in the Institute of Student Employers’ 2026 survey reporting it, up from 85% a year earlier and 78% two years ago. AI used without declaration in interviews is the most common form, reported by 66% of respondents, The Guardian reported on Wednesday 14 October. The same survey, read in full, shows employers moving in two directions at once: more of them now accept AI help with CVs and applications, while tightening up on its use in live assessment.

What recruiters see

Chris Morrow, whose Manchester agency Digitalent recruits for the AI sector, told The Guardian the signs are hard to miss in video interviews: candidates stall while a tool generates an answer, lean on fillers such as “That’s a good question”, and speak in unusually formal language. “It’s like they are reading off an Autocue,” he said. His agency ends the interview when it happens.

ISE, whose 290 members include Accenture, Rolls-Royce and Marks & Spencer, did not identify the employers affected. After interviews, the next most common misuse was in online assessments, reported by 47%, followed by application questions and then CVs and covering letters, at 10%. Chief executive Stephen Isherwood’s concern is practical: “If you misrepresent yourself, you could end up doing the wrong job.”

What the survey adds

ISE’s own summary of the Student Recruitment Report 2026 fills in the response. Employers offering guidance on acceptable AI use rose to 82%, from 55% a year earlier, and 64% now accept AI somewhere across hiring, up from 46%, mainly for writing CVs and cover letters. The red line sits at online assessments, interviews and assessment centres. Countermeasures include time limits, randomised question banks and more personal questions.

The jobs picture

The survey plays down fears that AI is already shrinking graduate hiring. After an 8% fall last year, the market rose 4% in 2026 and employers expect another 7% in the coming year, although 40% are cutting intake while 45% are increasing it. Isherwood attributes the cuts to economic uncertainty, not automation, and places the AI effect on graduate jobs “two or three years down the line”.

Looking forward

In our view the interesting number is not the 92% but the 82%: most employers have now written down what AI use they permit, which shifts “misconduct” from a judgement call to a policy breach. That mirrors the police watchdog’s message this week that clear rules protect the individual as much as the institution. For UK employers still without such guidance, the ISE figures suggest the gap will be filled by candidates first.