Candidates using AI prompts get caught mid-interview: 66% of employers have seen it firsthand

His agency, Digitalent, recruits for the AI sector, and he runs a steady stream of these interviews. An ISE report dated Oct 14, 2026 turns his case into a record: 92% of UK employers suspected or caught candidates cheating, up from 85% a year earlier and 78% two years earlier.

How a recruiter can tell a human isn't answering

Three signs. A pause while the tool puts the answer together. Fillers like "That's a good question". Overly formal speech, "as if they're reading from a teleprompter". Morrow ends the interview on the spot.

AI is used most often in the live conversation itself: 66% of employers reported such a case. In online assessments, 47% saw undeclared AI use, and in CVs, cover letters and application forms, 10%.

A study by the Fabric platform of 19,368 interviews (July 2025 to January 2026) found signs of prompting in 38.5% of candidates: 48% in engineering roles versus 12% in sales. Of those who weren't caught, 61% clear the hiring bar.

What is allowed now and what isn't

Almost everyone sets rules. A year ago, 55% of employers explained the boundaries to candidates; now 82% hand out guidance on where AI is and isn't allowed. Permissions are up too: 64% accept AI at some stages, versus 46% a year earlier, most often for the CV and cover letter. Live prompting during an interview and having AI complete assessments for the candidate remain off-limits.

Claire Tyler, who leads research at ISE, puts it plainly: find out a specific employer's rules before you turn on AI.

AI has also reached the other side of the table. 16% of employers already use it to screen applications, another 29% plan to within a year, and 77% are rethinking their selection process. Competition is tight: 141 applications per graduate vacancy on average, 172 in London.

Gartner expects that by 2028, one in four candidate profiles worldwide will be fake.