Ask most hiring managers what screening means and they will describe it as the boring middle step: the point between “post the job” and “interview the shortlist” where someone reads a pile of CVs and throws out the obviously unqualified ones. That description was always a little thin. In 2026, it is dangerously out of date.
A recent EY report analysing over one million pre-employment screenings across more than 90 Indian organisations found a sharp rise in employment fraud, and not among inexperienced freshers either. In healthcare, 96% of fraud incidents involved candidates who had previously held a job. In finance it was 88%, and in IT and ITeS, 79%. Screening was never just a filter for competence. It was always, quietly, a filter for honesty too. The difference now is that dishonesty has industrialised, and most screening processes have not caught up.
What Screening Actually Is, Once You Strip the Euphemism Away
Screening is the set of checks a business runs on a candidate before it commits real time, a formal interview, a reference check, sometimes a job offer, to finding out if they are worth that investment. It typically covers whether someone’s stated skills and experience match the role, whether their communication and basic fit come through in a short conversation, and increasingly, whether the person and the history on the page are actually real.
That last part used to be an afterthought, something a diligent recruiter did informally through a LinkedIn glance or a gut instinct on a phone call. A pre-employment testing survey found that up to 78% of resumes contain misleading statements, and 46% contain outright lies, which shows authenticity was never a fringe concern even before generative AI arrived. What has changed is the scale and the tooling available to both sides.
The Stages Everyone Lists, and the One Change That Matters Most
Most explainers walk through the same sequence: resume screening, phone screening, skills assessment, interview, background verification. It is a reasonable structure, and there is little to argue with in the basics. Resume screening narrows a large pool against the job specification. A phone screen tests communication, interest and basic salary alignment before anyone invests in a formal interview. Skills assessments, whether an aptitude test, a coding exercise or a case study, filter for competence in a way a conversation alone cannot. Background and reference checks confirm the history a candidate has claimed, ideally before an offer goes out rather than after.
Where this guide departs from most of the others covering this topic is the order. The conventional structure treats identity and authenticity verification as the last box to tick, somewhere after the interview, right before onboarding. Security-focused hiring guidance is increasingly clear that this is backwards: identity verification should move earlier in the funnel, not later, ideally at the application stage, because by the time a synthetic or misrepresented candidate reaches a final interview, a business has already spent real recruiter hours on someone who was never going to be real in the first place.
Why Screening Cannot Just Be About Resumes Anymore
The uncomfortable truth is that a well-crafted resume now proves almost nothing. Generative AI has made it trivial to produce a CV that reads as articulate, well-structured and perfectly keyword-matched to a job description, regardless of whether the experience behind it is real. One HR lead at a Pune fintech described receiving 3,400 applications for a single mid-level developer role overnight, the vast majority polished to an almost identical standard, which is less a hiring pipeline than a wall of noise.
The fraud does not stop at the page either. Deepfake interview fraud rose sharply through 2024 and 2025, with a meaningful share of companies unknowingly advancing candidates who used AI-generated audio or video to pass a live screening call. Checkr’s research puts identity fraud among new hires at 23% of companies already reporting an incident, and Gartner has projected that by 2028, roughly one in four candidate profiles globally will be fraudulent in some form. Most Indian employers have yet to fully internalise how quickly this shifted from an edge case to a structural risk sitting inside a perfectly normal hiring funnel.
The part that should worry businesses most is not the fraud itself, it is how often it is spotted and ignored anyway. The ACFE found that 21% of organisations that fell victim to employment fraud had proceeded with onboarding despite red flags surfacing during screening. That is not a technology failure. That is a process failure, someone saw a warning sign and a deadline won anyway.
What Good Screening Actually Looks Like Now
The businesses getting this right in 2026 have stopped treating screening as a single checkpoint and started treating it as a layered defence, with each layer catching what the previous one cannot. AI-assisted resume parsing and ranking still earns its place, it is genuinely the only realistic way to handle volume when a single posting can attract thousands of applications. But ranking for keyword fit and verifying that a person is who they claim to be are two different jobs, and conflating them is exactly how fraudulent but well-optimised CVs keep sailing through.
Cross-referencing a candidate’s claimed history against their LinkedIn, GitHub or other public professional footprint is no longer a nice-to-have step reserved for senior hires, it is a basic hygiene check worth doing earlier and more often than most SMEs currently manage. Structured interviews, with consistent questions asked of every candidate, make it easier to spot the small inconsistencies that a scripted or synthetic candidate cannot maintain under genuine follow-up pressure. Background verification belongs before an offer is signed, not as a formality that happens after, when unwinding a bad decision is far more expensive than preventing one.
The honest conclusion here is that AI should be compressing the slow, repetitive part of screening, sorting volume, flagging obvious mismatches, so that human judgement gets spent where it actually matters: reading the inconsistencies a machine is not yet built to notice.
What Getting This Wrong Actually Costs
None of this is abstract risk management. A bad hire can cost 30% to 50% of that person’s annual salary once recruitment, onboarding and lost productivity are accounted for, and that estimate assumes the candidate was simply a poor fit, not an outright fraudulent one, where the legal and reputational exposure runs considerably higher.
Most businesses still budget for the cost of a slow hire far more carefully than they budget for the cost of a wrong one, and that balance is overdue a correction. A screening process built only to move fast will keep producing shortlists that look impressive on paper and fall apart on the first serious background check, and by then, the cost has already moved from the recruitment budget to somewhere far harder to recover from.
This is precisely the shift shaping how CareerFit runs screening: AI doing the sorting so recruiters spend their time on verification and judgement, not the other way round. A shortlist is only as good as the honesty of what is on it, and that remains a far harder thing to automate than a keyword match.