No AI hiring law in the European Union, the United States or the United Kingdom requires an employer to check that a candidate is who they say they are. Employers are buying those checks anyway, out of their own budgets, and that makes this a change in standard rather than a compliance cycle. It changes what an executive’s record has to survive.
Start with the thing almost every article on this subject gets wrong. The AI hiring laws now arriving across the European Union, the United States and the United Kingdom do not oblige an employer to verify anybody. They govern what the employer’s own system may do to a candidate: bias auditing, advance notice, transparency, human oversight, a right to an explanation. Not one of them says establish who this person is, or confirm that the qualification is real.
The European Union’s AI Act is the clearest test of that, because it is the most far-reaching of these regimes and it has almost nothing to say about the thing employers are actually buying. It restricts inferring emotions in the workplace, subject to medical and safety exceptions. It treats remote biometric identification, the one-to-many kind that picks a face out of a crowd, as high-risk. And it then expressly carves out biometric verification: the one-to-one check that confirms a person is who they claim to be. That carve-out is exactly the candidate identity check. The Act neither requires it nor meaningfully restrains it, which is the point.
Meanwhile employers are spending materially more on exactly that. Not because a regulator asked, but because they have decided the cost of not knowing exceeds the cost of finding out. That distinction matters more than it sounds. A statutory deadline can move, and this year several of them did: the European Union’s high-risk obligations for employment AI were due on 2 August 2026 and did not arrive. A screening budget approved after a board has been shown a fraudulent hire does not move back.
The useful name for what follows is the verification era: the period in which the scarce commodity in hiring stops being a claim that impresses and becomes a claim that an independent party will confirm. Executive hiring verification is the least measured corner of that shift and the most misread. For a generation of executives trained to be more findable, more visible and more differentiated, the reversal is awkward. Discoverability was the constraint. Corroboration is becoming the constraint.
One qualification belongs here rather than at the end, because it governs everything below. Almost none of the evidence for this shift was collected at executive level. It comes from volume hiring, from company financial reporting, from employer surveys and from named employers describing their own practice. We could not identify a public dataset measuring verification intensity in executive appointments specifically, and this article does not invent one. What it does instead is set out what is measurable, say plainly where the measurement stops, and reason from there to the executive question, which is not whether a leader will be verified but what happens to a career record built for a market that mostly took claims on trust.
Employers are paying for verification themselves, and the money is the strongest evidence available
The clearest signal is not a survey. It is revenue.
First Advantage Corporation, one of the largest background screening companies in the world, reported second-quarter 2026 revenue of $448.8 million on 6 August 2026, up 14.9% year on year, with adjusted EBITDA of $128.5 million at a 28.6% margin and full-year guidance raised. Its first-quarter growth was 8.6%. The company attributed the quarter to new logo wins including 20 enterprise bookings, base growth, upsell and cross-sell, and to increased customer demand across a number of its verticals, including transportation and logistics, retail and e-commerce, industrials and manufacturing, and general staffing. Read it for what it is: commercial evidence consistent with rising demand for screening, from a company that also gained share. It is not a market census, and it is growing against a hiring market that is not growing at anything like that rate.
Intent points the same way. First Advantage’s 2026 Global Background Screening Trends Report, published 13 March 2026 from a survey of more than 5,000 HR leaders and job seekers across nine industries and five regions, reports that 89% of HR hiring managers plan to add background screening or identity verification within two years. That figure should be read with its publisher in mind, since First Advantage sells the product the finding recommends buying, and the base for the 89% is the hiring manager subset rather than the full 5,000, which the report does not disclose. HireRight’s 2026 Global Benchmark Report of 30 June 2026, drawing on more than 1,900 HR, risk and talent acquisition professionals surveyed between 18 February and 6 March 2026, reports the same direction without published percentages: employment verification was the check most likely to uncover inconsistencies globally, and the largest organisations reported the highest discrepancy rates.
Identity verification is also being built into mainstream screening as a product. Checkr, which states its platform is used by more than 120,000 businesses, launched an identity verification service on 4 March 2026 combining liveness detection, device and network intelligence and forensic document analysis, completing in roughly two minutes. Two minutes is the relevant number, because it changes what is affordable. A check that took a week was reserved for finalists. A check that takes two minutes can be run on everybody, and products get used the way their economics allow.
The most quotable evidence is testimony from employers who name their reason. Alphabet’s chief executive Sundar Pichai, reported by Axios on 12 August 2025, said Google would “introduce at least one round of in-person interviews for people, just to make sure the fundamentals are there”. Cisco’s vice president of global talent acquisition, Scott McGuckin, told Computerworld on 26 August 2025 that “remote work and advancements in AI have made it easier than ever for fake candidates to infiltrate the hiring process”, and that Cisco was increasing verification steps and enhanced background checks that may involve an in-person component. A McKinsey spokesperson in the same coverage described face-to-face interaction as necessary to assess “judgment, empathy, creativity, and connection”.
That is not confined to companies large enough to make news. In Greenhouse’s The 2026 AI in Hiring Report, published in 2026 and covering the United Kingdom, Ireland and Germany, 86% of 571 recruiters and hiring managers said they had caught or suspected candidate fraud in the previous 12 months, most commonly fake references, and 31% said they now conduct more in-person, on-site interviews. That report does not disclose its fieldwork dates, so it is dated here by its year of publication only. Gartner, in its Top Future of Work Trends for CHROs in 2026 of 12 January 2026, named the reversal of the candidate fraud arms race as one of its trends and recommended combining in-person interviews and experiential assessment with AI tools, which is an analyst position rather than a dataset and should be read as one.
What this amounts to is a standard changing rather than a rule being obeyed. Revenue reported to a securities regulator under penalty of civil liability is a different class of evidence from a survey response, and it points the same way. Employers are buying verification because they have decided the cost of not knowing exceeds the cost of finding out, and there is no regulator in that calculation at all. For an executive, the reasonable inference, and it is inference rather than measurement, is that the process which used to begin after an offer, quietly and as a formality, is moving forward into the assessment itself and becoming part of how the decision is made rather than a confirmation of one already taken.
AI hiring law was never a verification mandate, and it is loosening anyway
This is where most current writing on the subject is wrong, and it is wrong in a way that is checkable in an afternoon.
The European Union’s AI Act classifies recruitment, selection, promotion, termination, task allocation and performance monitoring as high-risk uses. What follows from that classification is a compliance regime aimed squarely at the employer’s system: risk management, data governance, technical documentation, logging, human oversight, accuracy, conformity assessment, registration, and a duty to tell workers and candidates that the system is being used on them. Nothing in it asks whether the candidate is who they say they are.
That regime was scheduled to apply from 2 August 2026. Under Regulation (EU) 2026/1744 of 8 July 2026, the Digital Omnibus on AI, it was postponed to 2 December 2027 for stand-alone high-risk systems and 2 August 2028 for high-risk AI embedded in products. Employment uses remain classified as high-risk. So the core duties for employment AI are 16 months further away than employers were planning for.
The rest of the Act is not dormant, and anyone telling you otherwise is overselling the delay. The Article 5 prohibitions have applied since February 2025, including the prohibition on inferring emotions in the workplace outside the medical and safety exceptions. The general-purpose AI rules have applied since August 2025. The transparency obligations arrived in August 2026. What was deferred is a specific set of chapters, not the statute.
Colorado’s SB 24-205, cited in a great deal of published commentary as the leading United States example, never took effect. It was postponed to 30 June 2026 and then superseded by SB 26-189, signed on 14 May 2026 and effective 1 January 2027, which requires advance notice, post-adverse-outcome disclosure within 30 days in plain language, meaningful human review, three years of records and updated vendor contracts. That is a materially different law with a later start, and any watch list still naming SB 24-205 as live is describing something that does not exist.
New York City’s Local Law 144 is the most-cited of the measures actually in force, and the only one we could find an independent enforcement audit for. It is not the only one: Illinois has regulated AI video interviews since 2020, Maryland has required consent for facial recognition in interviews since 2020, and California’s employment regulations on automated decision systems took effect in October 2025. But the audit is what makes New York instructive. The Office of the New York State Comptroller audited the Department of Consumer and Worker Protection’s enforcement of the law in a report dated 2 December 2025, covering July 2023 to June 2025. Over that two-year period the department received two complaints relating to automated employment decision tools. Its own review of 32 companies identified one instance of non-compliance. The Comptroller’s independent review of the same 32 companies identified at least 17 instances of potential non-compliance. The audit further found that the department lacked the technical expertise to evaluate automated employment decision tools and that its complaint routing process was ineffective.
The direction elsewhere is looser rather than tighter. Illinois HB 3773 took effect on 1 January 2026, amending the Illinois Human Rights Act to require employee notice and to affirm existing non-discrimination duties, though the Illinois Department of Human Rights temporarily withdrew its proposed implementing rules, leaving the operating detail unsettled. It does not require candidate verification. In the United Kingdom, the Data (Use and Access) Act 2025 took effect in relevant part on 5 February 2026 and replaced UK GDPR Article 22. The picture is a liberalisation rather than a removal, and the detail matters. A significant decision based entirely or partly on special category data still may not be taken solely by automated means unless a statutory condition is met. For every other significant solely-automated decision the old prohibition is gone, replaced by a safeguards regime: the controller must give the individual information about the decision, let them make representations, let them obtain human intervention, and let them contest it. Employment lawyers at Littler observed on 26 February 2026 that it is now significantly easier for employers to use automated decision-making in the United Kingdom than in the European Union. Easier is not unregulated, and an executive told a decision about them was made by a machine has a route to a human.
In the United States, Executive Order 14365 of 11 December 2025 directed an AI Litigation Task Force and a Commerce evaluation of onerous state AI laws, naming Colorado’s as an example. State laws remain in effect and preemption would require Congress, but the federal executive is now a party arguing against them rather than a force behind them.
Read together, these five jurisdictions dispose of the compliance explanation, and they dispose of it twice over. None of these laws asks an employer to verify a candidate in the first place. And the parts that do regulate hiring are being postponed in the European Union, restarted later in Colorado, under-enforced in New York City, left unsettled in Illinois and loosened in the United Kingdom. That is not employers preparing for a deadline. It is employers acting on risk they have priced themselves.
The executive consequence is the durability point, and it is the reason this distinction is worth an executive’s time rather than a compliance officer’s. Requirements imposed by regulation can be lobbied, delayed and diluted, as this year demonstrates at length. Standards adopted because a business decided they were cheaper than the alternative tend to persist and to spread through supply chains, because the organisation is not waiting for permission to keep them. An executive planning a move on the assumption that scrutiny will ease when the rules slip has read the situation backwards.
The regulation that does compel verification checks identity and credentials, not capability
There is real law forcing verification. It is simply not AI law, and it does something narrower than the coverage implies.
In the United Kingdom, section 48 of the Border Security, Asylum and Immigration Act 2025 comes into force on 1 October 2026, with a revised Code of Practice on preventing illegal working. It extends right to work check duties well beyond direct employment, to non-employee contracts, individual subcontractors, service providers engaged through online matching platforms, and arrangements involving substitution rights. It requires proportionate systems and processes to ensure that the individual performing the work is the individual who was checked, which may involve facial verification technology, access passes or biometric systems. Civil penalties run to £60,000 per worker, with criminal liability possible. Whether it reaches you depends on how you are engaged. An executive supplied under an onward contract, or placed through a matching platform, is within scope. Someone genuinely trading on their own account, contracting directly with a client, is expressly outside it. Seven weeks out, that is a distinction worth knowing which side of you sit.
In South Africa, the NQF Amendment Act 12 of 2019 commenced on 13 October 2023 on the South African Qualifications Authority’s own account, with the exception of clauses 1(h), 3(3) and section 32A(1). What is in force is the part with teeth: falsely claiming a qualification registered on the National Qualifications Framework is a criminal offence carrying a fine or up to five years’ imprisonment, and SAQA maintains a register of misrepresented qualifications and a register of fraudulent ones, alongside a register of professional designations. What is not in force is section 32A(1), the provision that would oblige employers and organs of state to authenticate a qualification before appointment and to refer unregistered ones to SAQA for verification. Worth knowing precisely, because SAQA’s own published explainer describes that duty as though it were already operating. The infrastructure behind it is now being built out. Business Day reported on 30 April 2026 that SAQA has removed 1,100 qualifications from the national register, is phasing out all pre-2009 qualifications by 2029, and has placed QR codes on certificates linked to its database so that an employer can confirm in real time whether a qualification is valid, what training was completed and over what period. SAQA’s chief operating officer Makhapa Makhafola told the paper that fabricated qualifications are a growing problem. That report is a single source, and the 1,100 figure and the 2029 date are worth confirming against SAQA directly before anyone relies on them.
What both instruments verify is worth stating precisely, because the distinction gets blurred constantly. They verify identity, status and credentials. They do not verify capability, judgement or whether a leader did what their record says they did. A machine-checkable certificate closes the easiest category of misrepresentation and leaves the hardest one untouched.
That gap is where senior hiring lives. Nobody has ever appointed a chief executive because their degree was authentic. The effect on an executive is second-order and inferred rather than measured: as the cheap checks become automatic and instant, attention and budget move to the expensive checks, which are the ones about conduct, scale and judgement. A market that can settle credentials in two minutes will spend the time it saves somewhere, and it will spend it on the claims that matter.
None of this evidence was collected at executive level, and the difference matters
Stating the limitation properly is more useful than working around it.
We could not identify a published figure for the proportion of executive hires subject to identity verification, an executive misrepresentation rate, or a dated measurement showing that referencing or due diligence at executive level became more intense between 2024 and 2026. In the public research of the Association of Executive Search and Leadership Consultants, the industry body for retained search, we found no quantitative benchmark covering verification, referencing or executive due diligence; its recent output concerns AI operating models for search firms. The absence is itself informative: the profession that does the most verification at senior level is not currently telling a verification story.
What exists at executive level is practitioner description. Morgan Taylor, a partner at the Mintz Group, told Hunt Scanlon Media on 22 April 2026 that traditional checks are narrowly focused on major potential issues such as criminal records or misrepresented degrees, and miss senior-level risk. The areas he describes as newly emphasised are reputational assessment through peer interviews and media analysis, historical review extending over 15 to 20 years or more, examination of litigation in which the candidate is referenced rather than named, leadership style and cultural impact, and pattern recognition in judgement across organisations. That is qualitative, it comes from a firm that sells due diligence, and it carries no figures at all. It describes a broadening rather than an arrival.
Two further facts narrow the claim honestly. Verification is not new at senior level; retained search has always involved referencing and board scrutiny, so anything happening now is an intensification of an existing practice rather than the introduction of a missing one. And most senior appointments never pass through the channel where the verification evidence was collected. Drawing on Spencer Stuart, Heidrick & Struggles and Crist|Kolder reporting, our analysis of the four channels that carry senior appointments and how each is entered found that 60% of S&P 1500 chief executive appointments in 2025 were internal promotions, that 63% of sitting chief executives were appointed internally, and that 85.3% of chief operating officer hires across 667 companies were internal. Internal candidates are far less exposed to the application-stage risks driving this activity, because the organisation already holds first-party employment and identity information about them, though senior internal appointments can still attract references, conflict checks and fit-and-proper processes.
So the honest form of the argument is this. The verification evidence is strongest in applied, high-volume hiring, and the applied route carries the smallest share of senior appointments. What travels upward is not the screening technology but the norm it establishes, and norms travel through the same organisations that set them: the board that approved the identity verification budget for graduate intake is the board that will ask, 18 months later, how thoroughly the incoming group finance director was checked. That is reasoning, not measurement, and it is presented here as reasoning.
What follows from it for a leader is a question of sequencing rather than belief. An executive who builds a record that would survive an identity check, a 20-year historical review and a peer interview with someone they did not nominate loses nothing if the shift reaches their level late, and is exposed if it has already arrived. The cost of preparing early is a few deliberate decisions about what gets documented and who gets asked; the cost of preparing late is that neither can be created inside the window of a live process.
The strongest objection is that this is rationing rather than verification
A senior reader will arrive at this objection quickly, and it is a good one.
Greenhouse’s Benchmark Report of March 2026, covering more than 6,000 North American organisations and 640 million applications from 2022 to 2025, reports that total applications per organisation rose 128.1% since 2022, reaching 41,422 a year, and that applications per role rose to 244 in 2025 from roughly 115 in 2022. Over the same period, average recruiting team size fell 55.6%, to five recruiters per organisation. Those are platform figures rather than a market-wide sample, and they describe a specific arithmetic: roughly twice the volume arriving at less than half the people.
An industry in that position does not look like an industry checking claims more carefully. It looks like an industry looking at far fewer candidates and checking almost none of them properly. Verification and triage produce similar surface behaviour, and from the outside they are difficult to tell apart.
Two further pieces of evidence sharpen the objection. SHRM’s 2026 Recruiting Executives Benchmarking: Attracting Critical Talent, surveying 4,657 members between 24 November 2025 and 23 January 2026, found that only 20% of organisations measure quality of hire, unchanged since 2025. A market that had genuinely become evidence-driven about candidates would be expected to become evidence-driven about outcomes, and on this measure it has not moved at all. And announced changes to hiring practice have a poor record of becoming practice: the Burning Glass Institute and the Harvard Business School Project on Managing the Future of Work found, in Skills-Based Hiring: The Long Road from Pronouncements to Practice of 14 February 2024, that firms announcing the removal of degree requirements largely did not change who they actually hired.
The objection does not fully hold, for one reason. Rationing and screening-budget growth are not alternatives; they are the same story from two ends. When the people reviewing applications more than halve while volume doubles, the human capacity to judge a claim collapses, and one of the things that replaces it is a purchased check. Automated ranking, triage and assessment tools are others, and they are not mutually exclusive. But purchased verification is the part with a revenue line attached to it, which is why First Advantage’s growth and Checkr’s two-minute identity product are visible in a way the rest is not. The rationing evidence explains why verification is being bought rather than performed.
Where the objection lands, and it lands hard, is on the quality-of-hire finding. An industry that verifies inputs and does not measure outputs has bought assurance rather than judgement. The executive implication is unattractive but worth knowing: if this norm keeps moving into senior hiring, leaders will face more extensive corroboration without any matching evidence that the process produces better appointments. Being checkable is now necessary. It is not sufficient, and nobody should present it as a route to being chosen.
Employers are demanding a verifiability they do not offer in return
The asymmetry in this market is the part almost nobody writes about, and it is the part an executive should find most useful.
Research by Resume Genius reported by Forbes on 16 July 2026, based on 1,500 United States hiring managers, found that 86% say AI makes it harder to verify candidate authenticity, while only 35% of companies always disclose their own use of AI in candidate evaluation and 20% never disclose it. The same population reports heavy AI use in hiring. Employers are asking candidates for a transparency they are not offering in return.
The detection gap underneath it is wider still. In iprospectcheck’s 2026 State of Screening Report, published 29 July 2026 from a survey of 1,500 United States business managers and owners with disclosed methodology, 14.0% reported encountering AI-generated CVs. In the 2026 Greenhouse report cited earlier, 78% of a 1,700-strong jobseeker panel across the United Kingdom, Ireland and Germany said they use AI to tailor CVs or application materials. Those are different countries, different populations and, more importantly, different questions: using AI to tailor a paragraph is not the same act as submitting what an employer would call an AI-generated CV. The honest conclusion is not that employers detect 14% of it. It is that the available studies define AI assistance too inconsistently to establish how reliably employers detect it at all, which is its own kind of answer. That is why the response takes the form of identity checks, in-person rounds and third-party verification rather than better reading: you cannot inspect your way out of a problem you cannot see.
Two figures from the same iprospectcheck survey show what employers do with that uncertainty. Fully 80.9% believe identity verification should be standard practice, and 30.2% have withdrawn an offer because of CV fraud. The second is the one that matters, because a withdrawn offer is a self-reported behaviour rather than an attitudinal preference, and it is the kind of event that gets a screening budget approved.
Now the surprise, which reverses the usual framing. Verification is normally presented as friction imposed on candidates. Candidates do not appear to read it that way. A second-quarter 2025 Gartner survey found 62% were more likely to apply where in-person interviews are required. A separate Gartner survey of 3,000 candidates in the same quarter found 6% admitting to interview fraud, either posing as someone else or having someone else pose as them. An earlier survey of 2,918 candidates found only 26% trust AI to evaluate them fairly, and only half believed the jobs they were applying for were legitimate. Gartner reports the 62% without explaining it.
One plausible reading is that a visible human stage signals organisational seriousness. In a market flooded with applications, automated evaluation and roles that may not exist, a demanding process is one of the few available signals that the counterparty is real. For an executive assessing an approach, that cuts both ways and is worth using deliberately. A process showing no credible attempt at corroboration at any stage is not a compliment to your reputation. Plenty of legitimate international searches run entirely on video, and references often come only at the final stage, so no single missing element proves anything. The absence of any of it, weighed alongside the provenance of the approach and the transparency of the mandate, is information about the seriousness of the brief, and it should be read alongside what a genuine route to a C-level appointment usually looks like.
The Corroboration Ladder: where a claim sits determines what it is now worth
Every claim in an executive record sits on one of four rungs, defined by who will confirm it. The rungs describe increasing independence of corroboration rather than universal probative strength, and different claims need different evidence: audited accounts prove the revenue rose and say nothing about who caused it, while a board paper may prove the second and not the first. The ladder is this article’s analytical device, derived from the verification practices described above rather than from any dataset of executive appointments, because no such dataset is published.
- Asserted. The only evidence is the executive’s own statement. Most CV achievement lines live here, including almost every claim beginning with “transformed”, “turned around” or “grew”.
- Documented. The executive holds a document supporting the claim: a certificate, a contract, a board paper, an internal deck. Stronger, but the document is in the claimant’s possession, which is the exact condition verification exists to get behind.
- Registered. An independent system confirms the claim without the executive’s participation. A QR-coded SAQA certificate, a professional body register, a companies registry, published financial statements, a regulatory filing, a court record, a published appointment announcement.
- Witnessed. A named person who was present will confirm it under questioning, and can be reached by somebody who did not get their name from the executive. This is what 360-degree referencing and the peer interviewing described by the Mintz Group reach for.
The evidence in this article supports one operating rule about that ladder. The verification era does not raise the bar on what an executive claims. It discounts the claims that only the executive can confirm. Nothing in the screening market, the regulatory position or the employer testimony above increases the value of a well-written rung one statement. Everything in it increases the value of rungs three and four, because those are the rungs a purchased check and an experienced consultant can actually reach.
Applied honestly, the ladder is uncomfortable, which is what makes it useful. Take the three achievements a leader would lead with in an interview and place each on a rung. For most executives, the headline financial numbers sit on rung one unless the company published them, the qualifications sit on rung three, the scale markers such as headcount, budget and geography sit on rung one or two, and the judgement claims, which are the ones the appointment actually turns on, usually sit on rung four, if they sit anywhere. Board papers, documented decisions, regulatory findings and formal assessments can carry some of that weight, but they are rarely in the executive’s own hands. That is not a document problem and it will not be solved by better phrasing. It is a question of which evidence exists and who is willing to speak to it, which is why the way an executive CV is structured for the readers it actually meets matters less than what those readers will find when they go looking.
Two practical consequences follow. Rung four evidence has to be cultivated years before it is needed, because a reference cannot be manufactured at the point of a process and a peer interview reaches people the executive did not nominate. And rung three evidence should be created deliberately while a leader is in post: published results, named roles in filings, industry body positions, on-the-record commentary and recorded appointments all convert private history into checkable record. That is a large part of what executive personal branding does and does not achieve, and it is a more defensible reason to do it than visibility for its own sake.
Why this article gives no figure for AI screening adoption and no CV fraud rate
Two of the most demanded numbers in this subject are missing from this piece deliberately, and the reasoning is worth more than the numbers would have been.
On AI screening adoption, credible sources with disclosed methodology do not agree, and the disagreement is structural rather than resolvable. SHRM’s State of AI in HR 2026, surveying 1,908 United States HR professionals between 5 and 23 December 2025, found recruiting to be the most common HR area using AI at 27% of organisations, with 39% using AI anywhere in HR and 31% having no plans to. iprospectcheck’s survey of 1,500 United States business managers and owners found 31.6% using AI somewhere in hiring. Research by Resume Genius reported by Forbes, from 1,500 United States hiring managers, found 87% of organisations using AI in hiring and 58% using it to screen CVs. The spread from 27% to 87% is a question-wording and population problem rather than a data problem. Any single headline percentage in this field has been selected from a range that wide, and several of the near-universal screening figures circulating most widely of all we traced to secondary career content for which we could not identify an underlying survey.
On CV fraud rates, the published estimates are too heterogeneous to support a defensible single number. The rates in circulation come from United States self-report panels, mostly published by companies selling CV products or screening services, and they disagree by a factor of five.
The distinction that resolves it is one this category almost never makes. Asking people whether they have lied on a CV and measuring what verification actually finds are different measurements of different things, and the first is worth very little. Only verification-outcome data with a published denominator is usable. The best such figure available for South Africa comes from MIE’s Background Screening Index covering 3.2 million screening transactions: 6.59% of 652,133 qualification verifications contained discrepancies, rising to about 11% for internationally issued qualifications, a subset whose denominator MIE does not disclose. That is 2024 data and it is the most recent MIE index published as at August 2026, which puts it outside the 12-month window this firm normally requires for market data, and it is stated here with that limitation attached rather than dressed up as current.
Executives should apply the same test to any hiring statistic put in front of them, including by a search firm or an adviser. Ask what the denominator was, who paid for the research and whether the number describes what people said or what somebody found.
The record has to be true and reachable before the writing matters
The shift is not that senior hiring has become suspicious. It is that the market’s tolerance for uncheckable claims has fallen at the same moment that uncheckable claims became free to produce, and the correction is being funded by employers rather than mandated by anybody. That makes it durable in a way a compliance cycle is not.
Five questions worth being able to answer about yourself:
- Of the three achievements you would lead with, how many can be confirmed by someone other than you, and by whom specifically?
- Which of your material claims exist on an independent register or in a published document, and which exist only in your own account of your career?
- If a due diligence firm reviewed 15 to 20 years of your record, including litigation in which you are referenced rather than named and commentary you did not author, what would it find that you have not addressed?
- Who would be interviewed about you by someone who did not get their name from you, and what would they say about your judgement rather than your results?
- When an approach arrives with no referencing and no attempt at corroboration at any stage, what does that tell you about the mandate behind it?
None of these are document questions, which is the point. Corroboration is assembled years earlier, by other people, in places a leader does not control, and by the time a process opens it either exists or it does not.
Frequently asked questions
Are AI hiring laws in force in 2026?
Yes, but unevenly, and none of the measures below requires an employer to verify a candidate. The European Union’s high-risk employment obligations under the AI Act were postponed from 2 August 2026 to 2 December 2027 by Regulation (EU) 2026/1744 of 8 July 2026, the Digital Omnibus on AI, while the Act’s prohibitions, general-purpose AI rules and transparency obligations are already live. Colorado’s SB 24-205 never took effect and its replacement, SB 26-189, starts on 1 January 2027. Illinois HB 3773 is in force from 1 January 2026 but its implementing rules were temporarily withdrawn. New York City’s Local Law 144 is in force, and a New York State Comptroller audit dated 2 December 2025 found two complaints in two years and at least 17 potential non-compliances the city’s own review had missed.
What is actually driving employers to verify candidates?
Cost and risk, funded internally. First Advantage reported second-quarter 2026 revenue of $448.8 million on 6 August 2026, up 14.9% year on year, with guidance raised, while hiring volumes are broadly flat. Named employers including Google and Cisco have stated publicly that they added in-person interview rounds and verification steps because AI has made candidate authenticity harder to establish.
Does the verification shift apply to executive hiring?
We could not identify a published measurement of it at executive level: no proportion of executive hires verified, no executive misrepresentation rate and no dated change in referencing intensity, and no quantitative benchmark in the public research of the industry body for retained executive search. The evidence in this area comes from volume hiring, employer surveys and practitioner commentary, and the executive case is reasoned from those rather than measured directly.
Which regulation genuinely requires employers to verify?
Immigration and qualifications law rather than AI law. In the United Kingdom, section 48 of the Border Security, Asylum and Immigration Act 2025 takes effect on 1 October 2026 and extends right to work checks to contractors, subcontractors and platform-engaged workers, with civil penalties up to £60,000 per worker. In South Africa, the NQF Amendment Act 12 of 2019 has made qualification misrepresentation a criminal offence since 13 October 2023. Both verify identity, status and credentials rather than capability.
Sources and further reading
- First Advantage Corporation, Q2 2026 results, 6 August 2026
- First Advantage, 2026 Global Background Screening Trends Report, 13 March 2026
- HireRight, 2026 Global Benchmark Report, 30 June 2026
- Greenhouse, The 2026 AI in Hiring Report, covering the United Kingdom, Ireland and Germany, 2026. Fieldwork dates not disclosed by the publisher
- Greenhouse, Benchmark Report, North America, March 2026
- iprospectcheck, 2026 State of Screening Report, 29 July 2026, updated 3 August 2026
- SHRM, 2026 Recruiting Executives Benchmarking: Attracting Critical Talent, fielded 24 November 2025 to 23 January 2026
- SHRM, The State of AI in HR 2026, fielded 5 to 23 December 2025
- Gartner, Top Future of Work Trends for CHROs in 2026, 12 January 2026, and candidate surveys of 31 July 2025
- Computerworld, 26 August 2025, and Axios, 12 August 2025, on the return of in-person interviews
- SiliconANGLE, 4 March 2026, on Checkr’s identity verification launch
- Office of the New York State Comptroller, Enforcement of Local Law 144: Automated Employment Decision Tools, 2 December 2025
- Regulation (EU) 2026/1744 of 8 July 2026 (the Digital Omnibus on AI), amending Regulation (EU) 2024/1689, OJ 24 July 2026, deferring the Annex III high-risk obligations to 2 December 2027 and product-embedded high-risk systems to 2 August 2028
- Regulation (EU) 2024/1689 (the AI Act), Article 5(1)(f) on emotion inference in the workplace, Annex III point 1(a) on remote biometric identification and its express exclusion of one-to-one biometric verification, and Annex III point 4 on employment uses
- Ogletree Deakins, 16 June 2026, on the anticipated delay to the EU AI Act’s employment obligations. Cited as commentary published ahead of the Regulation; the Regulation above is the authority for the dates
- Illinois Artificial Intelligence Video Interview Act, 820 ILCS 42, in force 1 January 2020; Maryland Labor and Employment Code s3-717, in force 1 October 2020; California Civil Rights Council regulations on automated-decision systems in employment, in force 1 October 2025
- Data (Use and Access) Act 2025, section 80, inserting Articles 22A to 22D into the UK GDPR, in force 5 February 2026 under SI 2026/82
- Home Office, draft Code of Practice on preventing illegal working: Right to Work Scheme for employers, on scope, on facial verification technology and access passes as examples, and on the exclusion of individuals in business on their own account contracting directly with a client
- National Qualifications Framework Act 67 of 2008 as amended, sections 13, 32A and 32B, including the section 32B(6) penalty of a fine or imprisonment not exceeding five years, and the uncommenced status of section 32A(1)
- Seyfarth Shaw, 22 May 2026, on Colorado SB 26-189
- Littler, 26 February 2026, on the UK Data (Use and Access) Act 2025
- White & Case, 15 January 2026, on Executive Order 14365
- Lewis Silkin, 2 July 2026, on right to work changes from 1 October 2026
- SAQA, The NQF Amendment Act of 2019 and its Implications, 3 November 2023, and Business Day, 30 April 2026
- MIE, Background Screening Index, 2024, the most recent published as at August 2026
- Hunt Scanlon Media, 22 April 2026, on executive due diligence practice
- Forbes, 16 July 2026, reporting Resume Genius research on employer AI use and disclosure
- Burning Glass Institute and Harvard Business School Project on Managing the Future of Work, Skills-Based Hiring: The Long Road from Pronouncements to Practice, 14 February 2024
Last reviewed 14 August 2026.
Working with Elite Executive Career Solutions
Building a record that survives this kind of scrutiny is the work of the Executive Positioning Strategy engagement, which examines where a leader’s evidence actually sits, what a search consultant or due diligence firm would find, and what needs to be built before a process rather than during one. Where the work is documentary, it runs through the Executive CV & Cover Letter service. The full range is set out under Executive Solutions, the individual engagements are described at our executive engagements, and a confidential conversation can be arranged at [email protected].