How AI-generated documents are forcing recruiters to rethink what counts as proof.
The problem is no longer occasional — it’s systemic
For years, CV embellishment was treated as a minor irritant in recruitment: a stretched job title here, a rounded-up qualification there. That era is over. Across executive search, RPO, and direct recruitment, employment fraud, payslip fraud, and qualification fraud have become a recognisable and growing pattern — not isolated incidents, but a phenomenon recruiters and hiring managers now have to actively defend against on almost every senior mandate. Reference checks have never mattered more.
What has changed the game entirely is artificial intelligence. Candidates no longer need design skills or inside knowledge of payroll systems to produce convincing fake documentation. AI tools can now generate payslips with plausible company branding, tax reference formats, and salary structures. They can produce qualification certificates, transcripts, and even correspondence that mimics a university’s tone and letterhead. The barrier to producing fraudulent “proof” has collapsed — and with it, the reliability of documents that recruiters have leaned on for decades as a baseline standard of trust.
This is not a future risk. It is a present, operational reality for anyone conducting due diligence on candidates today.
Why documents alone are no longer proof
Historically, due diligence has rested on three pillars: the CV, supporting documents (payslips, certificates, ID), and reference checks. AI has quietly compromised the middle pillar. A payslip is no longer inherently trustworthy just because it looks correct — “looking correct” is now trivially achievable. The same applies to degree certificates and transcripts, particularly from institutions whose verification processes are slow, manual, or difficult to access across borders.
This doesn’t mean documents should be discarded. It means they must be treated as a starting point for verification, not the endpoint of it. Every document should be corroborated independently — through accredited verification platforms, direct institutional contact, and, critically, through people who can speak to the candidate’s actual conduct and performance.
Why reference checks are the one truth AI cannot fabricate
A CV can be written. A payslip can be generated. A certificate can be forged with unsettling precision. What is far harder to fabricate at scale is a genuine, unprompted conversation with someone who actually worked alongside the candidate — a former manager, peer, or direct report who can describe specific projects, specific behaviours, and specific outcomes.
This is why reference checks deserve to be elevated from a routine box-ticking exercise to the central pillar of due diligence. A flawed reference process — relying only on HR-issued confirmation letters, accepting references without verifying the referee’s identity and relationship to the candidate, or using generic templated questions — leaves exactly the gap that fraudulent candidates are counting on.
A rigorous approach to reference checks should:
- Verify the referee independently. Confirm the referee’s identity, role, and relationship to the candidate through a separate channel — not only the contact details the candidate supplied.
- Ask specific, open-ended questions that require detailed, first-hand recall rather than yes/no confirmation. Generic answers to generic questions are a red flag in themselves.
- Triangulate across multiple referees where possible, rather than relying on a single account.
- Probe employment dates, reporting lines, and reasons for leaving directly with the referee, and compare this against what the candidate has stated.
- Be conducted by someone trained to notice inconsistency — hesitation, vagueness, or answers that don’t match the specifics of the role being discussed.
Verification platforms remain essential — but only the legitimate ones
Alongside references, formal verification through accredited platforms and institutions remains a non-negotiable part of the process — qualification verification through recognised bodies, criminal and credit checks through registered providers, and employment history confirmation through legitimate payroll or HR systems rather than candidate-supplied documents alone. The key word is legitimate: as fraud has evolved, so has the sophistication of fake “verification” services designed to give false confidence. Due diligence partners and internal teams need to know precisely which verification channels are accredited and trustworthy, and should never substitute a document check for a proper verification query submitted directly to the source.
What this means for recruiters, HR teams, and hiring managers
The organisations that will protect themselves best in this environment are not the ones with the most paperwork — they’re the ones with the most disciplined process.
That means:
- Treating every supporting document as unverified until independently confirmed.
- Investing time and resourcing in genuinely thorough reference checks, not a rushed formality at the final stage of the hiring decision.
- Using only accredited, legitimate verification platforms for qualifications, criminal checks, and credit checks.
- Training recruiters and hiring managers to recognise the signs of AI-generated documentation and inconsistent referee accounts.
- Building due diligence into the process early, not as a last-minute gate before an offer is signed.
The bottom line
AI has made it easier than ever to fabricate the appearance of credibility. It has not made it easier to fabricate genuine human testimony from someone who actually worked with the candidate. That is precisely why reference checks — done properly, done rigorously, and done without shortcuts — remain the closest thing recruitment has to ground truth. In a landscape where a payslip or certificate can no longer be taken at face value, the conversations we have with the people who actually know a candidate’s work are more valuable than they have ever been.
Recruitment as an industry needs to recognise this shift and respond to it — not with more documents, but with better questions, better verification discipline, and reference checks built to withstand the very fraud it now has to detect.
Lorraine Barnard is CEO and Director of Dananda Global Talent.
How rigorous is your organisation’s reference-checking process — and would it withstand an AI-generated fraud attempt?
