AI in job applications: where does acceptable help become cheating?

Tom spent months chasing jobs after graduation, firing off applications by the dozen each day. Over 200 attempts before landing a restaurant manager role last month. The process wore him down—employers ghosted him, rejections piled up, and the emotional toll was real. He couldn't have survived it, he says, without help from artificial intelligence.

He used AI to polish his cover letters and CVs, to hunt for the keywords that automated screening systems flag and pass along to human eyes. He never let it write his video interview answers or complete his assessments. But he watched the contradiction unfold: companies demanded authenticity while simultaneously using their own AI to filter candidates before anyone read a single word. "You can't use AI too much because otherwise you're not human," he says. "But if you don't write like an AI, you may get skipped by the AI screening."

This tension is now baked into graduate recruitment across the UK. A summer survey by the Institute of Student Employers, which polled 155 of Britain's largest graduate employers, found something striking: 92 percent had encountered or suspected AI misconduct in their hiring process, up from 85 percent the year before. The most accepted uses—writing cover letters and CVs, completing application questions—sit in a gray zone that most employers tolerate. But the line hardens fast. Using AI as a live assistant during video interviews, or to answer assessments in real time, crosses into territory employers call cheating.

The pressure is relentless. Each graduate role now attracts an average of 141 applications. AI has turbocharged what the industry calls "spray and pray"—blasting out dozens of customized applications in a single afternoon, something that would have taken weeks by hand. Stephen Isherwood, joint chief executive of the ISE, says this is "driving up the competition and pressure on students" in a market already brutal by design. One recently employed trainee accountant told the BBC he'd applied for 50 graduate jobs and used AI in every single one. He kept it to research and CV refinement, he said, but he knew others who didn't. "After a while they become desperate," he explained. "If they see everyone else is using AI, they think how can I compete?"

Hollie Moody, head of early careers recruitment at EDF, the energy company, doesn't blame graduates for reaching for the tool. The jobs market is punishing. But she sees the damage accumulating. EDF opened its graduate scheme three weeks ago and closed it early after receiving 12,000 applications—a 10 percent jump from the previous year. AI almost certainly drove that surge. The company now issues guidance on acceptable AI use and tries to make its process "AI-proof" by reading CVs only in later stages and using skills-based assessments and video interviews to measure what candidates can actually do. Still, applicants find ways around it. "A lot of the answers are the same," Moody says. "Or people might pause whilst they wait for AI to come back to them with an answer. It's almost like people are reading a script."

The irony cuts both ways. Only 16 percent of employers surveyed currently use AI to screen applications, but more than half plan to introduce it within the near term, driven by sheer volume. EDF hasn't adopted it yet, but Moody acknowledges they may have no choice. What emerges is a kind of arms race: candidates deploy AI to compete, employers deploy AI to filter, and the gap between acceptable help and outright deception keeps shifting. Moody calls it a "moral dilemma" that employers and employees will keep wrestling with. But she insists on one point: AI cannot replace what her firm actually needs. "It's about authenticity, but also personality. We want diversity of thought in the business." The question now is whether a system built on volume and speed, where both sides are reaching for the same tool, can still deliver either.