There's something happening in the hiring world right now that not enough people are talking about. And if you're recruiting, you need to know about it.
AI isn't just helping people write better CVs anymore. It's applying for jobs on their behalf, tailoring applications in seconds, and in some cases, sitting in on interviews for them. Yes, really.
The hiring landscape has shifted significantly in the last 18 months and understanding what's out there is the first step to protecting yourself. So let's get into it.
Let's start with the application stage. There's a whole category of tools now, platforms like LazyApply, LoopCV, AIApply, JobCopilot and Sonara, that allow a candidate to upload their CV once and then sit back while the software applies for hundreds of jobs on their behalf. Every single day. Automatically.
We're not talking about a few applications either. LazyApply markets itself as being able to blast a CV across LinkedIn and Indeed with one click. AIApply claims it can submit 100 applications in under 48 hours. LoopCV runs continuously in the background, finding new postings and submitting tailored applications without the candidate lifting a finger.
And here's the bit that should really make you pause. Most of these tools don't just send the same CV to everyone. They use AI to tailor each application to the job description, rewriting the CV and cover letter to match the role's keywords, requirements and language. So when you read an application that seems like a perfect fit, it might not be a person who spent an hour crafting a thoughtful response. It might be software that did it in 30 seconds.
Some of these platforms go even further. They optimise CVs specifically to pass ATS systems (those automated filters a lot of businesses use to sift applications before a human ever sees them). Which means you may not even be seeing the best candidates at the top of your pile. You might be seeing the ones whose AI played the system best.
If auto-applying was the whole story, that would be challenging enough. But it's not.
There are now tools specifically designed to assist candidates during the interview itself. Software runs invisibly in the background of a video call, listens to the questions being asked, and feeds the candidate suggested answers in real time through a hidden window on a second screen. The interviewer sees nothing.
And then there's the deepfake side of things; software that manipulates someone's face and voice in real time on a video call. We know this is being used in interviews because it's already happened.
In 2024, KnowBe4, a well-known US cybersecurity firm, admitted they had hired a North Korean IT worker who used a stolen US identity and an AI-generated profile photo to get through their entire process. He passed background checks. He passed four video interviews. He got the job, received a laptop, and within hours began loading malware onto it. Their security team caught it within 25 minutes.
KnowBe4 is a cybersecurity firm. These are people whose entire job is spotting exactly this kind of thing. Their point in going public was simple: if it happened to them, it can happen to anyone.
The Cluely story is closer to what most employers will actually encounter. In early 2025, a 21-year-old Columbia University student named Roy Lee built a tool that ran invisibly during a live video interview, listened to the questions being asked, and fed the candidate AI-generated answers through a hidden overlay on their screen. The interviewer saw nothing. The candidate just read the answers.
Lee used it to land internship offers from Amazon, Meta and TikTok. Amazon pulled the offer. Columbia suspended him. He rebranded the tool as Cluely, launched it in April 2025 with the tagline "Cheat on Everything," raised $5.3 million in seed funding within days, and by June had secured a $15 million Series A from Andreessen Horowitz. Over 70,000 people signed up in the first week.
It sits invisibly on a candidate's screen. It listens to the interviewer. It generates answers in real time. It is designed specifically to be undetectable during screen sharing. This is not some obscure underground tool - it's a fully funded startup that went mainstream in 2025.
Closer to home, Deloitte UK reinstated in-person final stage interviews for their graduate and apprenticeship programme in September 2024 after the Financial Reporting Council flagged serious risks with their online process, reporting instances of AI-assisted cheating in assessments and calling it "unacceptable." One of the biggest names in UK professional services had to change how they hire because of this.
UK recruiters are regularly describing the same thing: candidates with weirdly polished answers, eyes drifting slightly off screen, who completely fall apart the moment you ask a follow-up question that wasn't in the script. One hiring manager put it well - the answers were "filled with buzzwords but with no substance whatsoever, and the speaking cadence of someone reading from a script."
An analysis of nearly 20,000 interviews by hiring platform Fabric found that 38.5% showed signs of AI-assisted cheating, with the rate tripling in late 2025. Critically, 61% of those candidates would still have passed a standard hiring process with no detection in place. They'd have got the job. And you'd have found out the hard way.
Here's where we want to be straight with you rather than just alarmist. The vast majority of people applying for your roles aren't North Korean operatives. Most aren't using deepfakes. But a growing number are using AI to get further than their actual skills and experience would take them, and that's the version of this problem most likely to land on your desk.
Think about what that means in practice. You spend time shortlisting, interviewing and onboarding someone who came across brilliantly. They start. And within a few weeks you realise the person in front of you isn't quite the person you met on that video call. The CV was polished by AI to hit every keyword. The interview answers were AI-coached to tick every competency. And now you've got someone in a role they're not equipped to do, and you're back to square one, only weeks later with a bad experience on both sides.
So what can you actually do about it?
Watch for the tells during video interviews. Slight delays before answering, eyes drifting slightly off-screen, an unnaturally polished cadence, or a perfect opening answer that completely falls apart when you follow up. If they can't build on what they just said, that's your signal.
Push for real, specific experience. "Tell me about a specific time when you..." is much harder to fake than "What would you do if...". Ask for timelines, what went wrong, what happened next. Real experience has texture. AI-generated answers tend to be suspiciously smooth.
Don't rely solely on remote interviews for important hires. Google reintroduced mandatory in-person interviews in 2025 specifically because of AI fraud concerns. For senior or high-trust roles, being in the same room still matters.
Be sceptical of applications that seem too perfectly tailored. Good candidates write good applications. But if a cover letter reads like it was engineered to hit every single point in your job spec in order, it probably was.
We don't use AI to screen candidates. Not because we're against technology, but because the screening stage is exactly where the human bit matters most.
When you work with us, every candidate on your shortlist has had a proper conversation with one of our consultants before you ever meet them. Not a form, not an automated assessment. A real conversation where we're actively trying to understand who they are, what they've actually done, and whether they're who they say they are.
We meet face to face where possible. We dig into the experience on the CV. We notice when answers feel rehearsed versus when someone is drawing on real memory. We ask the follow-up question, and then the one after that. Every placement we make is backed by a guarantee - three months as standard, six months if you go retained - because we stand behind the work we do.
We specialise in HR, Sales and Marketing, Finance, IT, Engineering and Executive appointments. These are roles where the gap between someone who can talk the talk in an interview and someone who can actually do the job is costly. Getting it wrong is disruptive. We're here to make sure you don't.
In a world where a tool can apply to 1,000 jobs overnight and tailor every single one to the job description, the filter that actually matters isn't the automated one. It's the human one.
That's what we do. And right now, it matters more than ever.