From AI-generated CVs to real-time interview assistance, the hiring process is becoming easier to perform and harder to trust. The organisations that succeed will be the ones that can still tell genuine capability from a well-rehearsed version of it.
One in four candidates applying for your next role may not be who they say they are.
Gartner now predicts that by 2028, 25% of candidate profiles globally could be fake — driven by the rapid rise of generative AI and identity manipulation in hiring. Now, the modern challenge in recruitment isn't just finding good people. It's knowing whether the person you're assessing is real, and whether their performance in the process reflects genuine capability or something carefully engineered behind a screen.
A recent Harvard Business Review article, AI Has Broken Hiring. Here's How to Fix It, makes a similar point. The early hiring funnel is weakening at both ends. At the top, CVs and cover letters can be produced, refined and tailored at scale. Further down, live interviews - long considered the most reliable signal of authenticity - are increasingly vulnerable to real-time assistance tools.
The cost of getting it wrong is rising
Hiring errors have always been expensive. But as it becomes easier to game the process, the stakes are climbing.
SHRM's 2025 benchmarking data puts average cost-per-hire at $5,475 for non-executive roles and $35,879 for executive hires. Gallup estimates that replacing an employee can cost between one-half and two times their annual salary once productivity loss, onboarding and disruption are factored in. And some estimates put the total figure higher still.
But the concern in the HBR research goes deeper than direct costs. When hiring processes begin to reward polish over substance - or familiarity with tools over underlying capability - organisations risk systematically selecting for the wrong qualities. Over time, that erodes team performance, capability building and organisational trust.
The problem isn't AI… it's gaming the process
AI itself isn't the issue. Used well, it can improve efficiency and support better decision-making. The challenge is that many of the ways organisations traditionally assess candidates are now easier to manipulate.
CVs are the obvious example. Research from the University of Maryland, the National University of Singapore and Ohio State has shown that AI screening tools can demonstrate self-preference bias, favouring CVs that resemble their own outputs. In simulated hiring pipelines, candidates whose applications matched the evaluating model's style were 23% to 60% more likely to be shortlisted than equally qualified candidates with human-written CVs.
At the interview stage, things get harder. Tools designed to provide real-time prompts or suggested answers during video calls are becoming more sophisticated and less visible. One such platform raised $5.3 million in seed funding with the explicit aim of supporting users during high-stakes interactions - including interviews.
Why human evaluation matters more now
Industrial and organisational psychology has long shown that structured, job-relevant interviews are among the stronger predictors of job performance - particularly when applied consistently and focused on reasoning rather than rehearsed responses. That finding holds up well under scrutiny, even as other assessment methods come and go.
What's changing is how those interviews need to be run. Static, predictable formats are easier to script against. More adaptive, conversational approaches - where interviewers probe, challenge assumptions and move off-script - are harder to replicate with real-time assistance and tend to surface more reliable signals of genuine judgment.
That shift places greater importance on the quality of the interviewer, not just the structure of the process.
Restoring credibility earlier in the funnel
One practical response is to establish authenticity earlier, before significant time and cost are invested further down the line.
This is where approaches like consultant-led video shortlists, such as LeonidLive, can play a meaningful role. When done properly, they offer more than a summary of experience. They provide an additional layer of visibility into how a candidate communicates, how comfortable they are discussing their own work, and how their presence aligns with the narrative on paper.
The value isn't in the format itself, but in the assessment behind it. When those conversations are conducted by experienced consultants rather than automated tools, they can test consistency, explore detail and surface gaps that written applications don't reveal. They can also bring forward candidates who might otherwise be overlooked.
When traditional indicators become less reliable, organisations often retreat into narrower networks; relying more heavily on known profiles and established credentials. That can reduce noise, but it also limits access to candidates whose CVs are less conventional but whose reasoning and judgment are strong. A more human-led evaluation process helps counterbalance that effect.
From efficiency to judgment
The issues AI is causing does not mean organisations should step away from technology. The trick is understanding which parts of the process can be automated safely, and which rely on human interpretation.
As the HBR article puts it: when competence becomes easier to simulate, judgment becomes what truly sets candidates apart.
For hiring leaders, the effectiveness of a process will depend less on how efficiently it filters candidates, and more on how reliably it distinguishes between those who perform well in the process and those who will perform well in the role.
That distinction has, of course, always mattered. It's just becoming harder to see…and more important to get right.
This is the first in our 'AI in Recruitment' thought leadership series....more coming soon...