37% of employers now rate CVs as a dependable indicator of talent (Willo Hiring Trends Report 2026, 2.5m interviews)
76% of hiring professionals encountered AI-generated applications in H1 2026, up from 53% in H1 2024
41% of enterprises have hired and onboarded a fraudulent candidate (GetReal Security Survey 2026)
The Structural Shift
The CV’s collapse as a hiring signal has been faster than most expected.
For most of its history, the CV served a practical purpose. It was an imperfect document, but it worked within a world where application volumes were manageable, content was human-generated, and a recruiter’s judgment could stretch across a realistic number of profiles. Those conditions have all changed at once.
Application volumes have grown by as much as 239% across almost every role category since the widespread adoption of generative AI tools, according to The Economist (2026). That growth has not come with any corresponding increase in signal quality. If anything, the opposite. The same tools that make it easy to apply to 200 roles in an afternoon also make it easy to produce a polished, keyword-rich CV that bears little relationship to what the applicant can actually do.
Source: The Economist (2026); LinkedIn Global Talent Trends Report (2026)
The consequences are now visible in employer behaviour. Research published this year by hiring platform Willo, drawing on data from over 2.5 million candidate interviews and input from major employers including Toyota and Microsoft, found that just 37% of hiring professionals now rate credentials and learning history as a dependable indicator of talent. Four in ten employers are actively moving away from CV-first hiring. Ten percent have replaced CVs with alternative evaluation methods entirely.
Source: Willo Hiring Trends Report 2026 - 2.5 million candidate interviews; employer panel including Toyota, Microsoft
“The CV used to tell a story of effort, experience, and aptitude. Now it often tells us how well someone can prompt a large language model.” — Euan Cameron, CEO, Willo (2026)
What is particularly significant is the speed of this shift. In 2024, CVs were still the default reference point for the majority of hiring decisions. By mid-2026, that default is breaking down across sectors. The employers responding most effectively are not those who have added more layers of CV screening. They are those who have reduced their dependence on the CV as a primary signal and replaced it with structured, evidence-based assessment at an earlier stage in the process.
This shift also has a fairness dimension that deserves attention. AI-polished CVs advantage candidates who know how to use AI tools effectively, not necessarily those who can do the work. A structured assessment process applied consistently to every applicant is, paradoxically, more equitable than a CV-first process that has always rewarded presentation over substance. The organisations making this transition are finding that their shortlists look different, and that the hires they make through structured evaluation perform better over time.
Source: Willo Hiring Trends Report 2026 - 70% of employers now use structured interviews; 73% rate their process as fair and inclusive
Official Warnings
Candidate fraud has become an enforcement priority, not just an operational inconvenience.
The deterioration of the CV as a reliable document has a more acute dimension than AI-assisted exaggeration. Across the United States and increasingly in Europe, hiring processes are being actively targeted by organised fraud operations using AI-generated identities, deepfake video technology, and proxy candidates to obtain employment under false pretences.
This is not a fringe risk. The FBI’s Internet Crime Complaint Center has issued multiple official public service announcements warning US businesses directly. The IC3’s guidance states that operatives are observed using AI and face-swapping technology during video job interviews to conceal their true identities. These candidates pass initial screening, clear background checks, and begin work, at which point they either perform minimally while collecting salaries that are funnelled to sanctioned foreign governments, or extract proprietary data for malicious purposes.
Source: FBI Internet Crime Complaint Center - PSA250723-4, July 2025 (ic3.gov)
The scale of documented enforcement is significant. In June 2025, the US Department of Justice announced coordinated actions against one scheme alone: two indictments, searches of 29 laptop farms across 16 states, and the seizure of 29 financial accounts used to launder proceeds. DOJ alleged that operatives obtained employment at more than 100 US companies using stolen or fabricated identities. Amazon’s Chief Security Officer disclosed separately that the company had blocked over 1,800 suspected North Korean applicants since April 2024, with attempts growing 27% quarter-over-quarter.
Source: US Department of Justice press release, June 30 2025; Crowell & Moring client alert, September 2025
Mandiant’s Chief Technology Officer stated publicly in 2025: “Literally every Fortune 500 company has at least dozens, if not hundreds of applications from North Korean IT workers. Nearly every CIO I have spoken to has admitted they have hired at least one.” The corporate victims documented in DOJ prosecutions total 479 to date.
Source: Mandiant (Google Cloud) - Security Conference Media Briefing, May 2025; DOJ prosecution records
What makes this particularly difficult for HR and talent teams is that the primary defence, the CV and video interview, are the two inputs most susceptible to manipulation. A CV can be fabricated in minutes using any large language model. A video interview can be conducted using real-time face-swapping software that passes visual inspection without specialist detection tools. The standard hiring process was not designed to defend against this. And most organisations have not updated it.
Gartner projects that one in four candidate profiles globally will be fake by 2028. Experian’s 2026 Future of Fraud Forecast identifies AI-driven employment fraud as a top-tier commercial risk, predicting that employers will unknowingly onboard false candidates at significantly larger scale in the next twelve months. The implication for HR leaders is direct: candidate verification is no longer a background screening task that sits at the end of the process. It needs to be built into the assessment itself, from the first substantive interaction.
Source: Gartner (2026) - Candidate Identity Fraud Forecast; Experian 2026 Future of Fraud Forecast
Regulatory Context
The regulatory direction is clear. The timeline just became more complicated.
The EU AI Act, which entered into force in August 2024, is unambiguous on the status of AI used in hiring. Under Annex III of the legislation, AI systems intended to analyse and filter job applications, place targeted job advertisements, or evaluate candidates are explicitly classified as high-risk. This is not a matter of interpretation, it is the text of the law. High-risk classification carries with it binding obligations: mandatory risk assessments, technical documentation, bias testing, human oversight mechanisms, transparency disclosures, and continuous monitoring.
Source: EU AI Act, Regulation (EU) 2024/1689, Annex III Category 4 - Employment and workers management
The compliance picture changed materially on 7 May 2026, when EU lawmakers reached political agreement on revisions to the AI Act under the Digital Omnibus package. A 16-month postponement now applies to new or substantially modified Annex III systems, which includes all employment-related AI. Rather than a fixed August 2026 deadline, obligations will now become applicable six or twelve months after the Commission confirms the availability of harmonised technical standards, depending on system category. In the absence of such a decision, the latest possible dates shift to December 2027 or August 2028. This agreement is still subject to formal adoption.
Source: Travers Smith - EU agrees to delay key AI Act compliance deadlines, May 7 2026
The postponement shifts the timetable. It does not change the direction. The obligations, documented decision logic, bias testing, human oversight, and audit trails, are coming regardless. And Article 26(7) already requires employers to inform and consult employee representatives before deploying high-risk AI.
The US picture is more fragmented but no less consequential. In January 2025, the EEOC removed its 2023 AI hiring guidance from its website following a change in administration. However, the EEOC’s Strategic Enforcement Plan for FY 2024-2028, which explicitly identifies technology-related employment discrimination, including algorithmic decision-making and AI, as a named enforcement priority, remains in force. It can only be modified by a quorum vote of commissioners. The removal of the guidance did not remove the underlying obligations. Title VII, the ADEA, and the ADA still apply. The guidance explained them; it did not create them.
Source: National Law Review - The Federal Government Quietly Removed Its AI Hiring Guidance, March 2026
Litigation is moving faster than regulation in the United States. In March 2026, a federal judge allowed age discrimination claims against Workday to proceed under the Age Discrimination in Employment Act, finding it legally plausible that an AI vendor could be liable as an employment agent under federal law. A parallel case involving Eightfold AI, which allegedly built reports on candidates without their knowledge using unverified third-party data, is also advancing. The key question courts are beginning to answer is not whether AI tools can produce discriminatory outcomes, but who bears liability when they do. The answer appears to be: both the vendor and the employer.
Source: Mobley v. Workday Inc., N.D. Cal. - March 2026 ruling; HR Dive, March 30 2026; Law360 Employment Authority, January 2026
At the state level, Colorado’s AI Act (SB 24-205) requires employers deploying high-risk AI systems to take reasonable care to protect consumers from algorithmic discrimination. California’s Civil Rights Council has extended anti-discrimination laws to AI tools, mandating four-year retention of automated decision data. New York City’s Local Law 144 requires independent annual bias audits for automated employment decision tools used in roles based in the city. These are no longer theoretical future requirements. They are operative now.
Source: Colorado SB 24-205 (eff. June 2026); California Civil Rights Council regulations (2025); NYC Local Law 144 (eff. 2023)
What Good Looks Like
The organisations handling this well share three characteristics.
The first is that they have reduced their dependence on the CV as a primary screening mechanism. This does not mean abandoning CVs entirely, for some roles and seniority levels they retain value as context. But it means not allowing CV quality to be the primary determinant of who advances. The CV tells you what a candidate claims. It does not tell you what they can do.
Research from Willo’s 2026 Hiring Trends Report found that 68% of employers now identify live behavioural interviews as the most trusted indicator of talent, up significantly from prior years. Hands-on skills demonstration and real-time problem-solving rank second and third. Close to 70% of employers now use structured interviews as standard. This is not a philosophical preference. It is a direct operational response to a hiring environment in which the CV’s signal value has degraded faster than most processes have adapted.
Source: Willo Hiring Trends Report 2026 - 68% identify live behavioural interviews as top talent indicator
The organisations performing best are not those with the most advanced AI. They are those that have integrated AI into a structured, evidence-led process, where AI supports human judgment rather than substituting for it.
The second characteristic is that their assessment process is designed to surface genuine knowledge, not optimised presentation. A candidate can refine a CV in minutes using any AI writing tool. They can prepare rehearsed answers to standard interview questions. What they cannot easily do is answer adaptive follow-up questions in real time about the specific work they claim to have done. A question that begins with the candidate’s own CV and then probes in depth, asking what they would have done differently at a particular point, what the failure mode was, how they handled the decision, either reveals genuine knowledge or it does not.
The third characteristic is documentation. Not documentation as a compliance burden, but documentation as a structural feature of the process. Every shortlisting decision recorded with reasoning. Every evaluation scored against consistent criteria. Every assessment producing an exportable record. This matters for three reasons: it protects the organisation in the event of a challenge or audit; it reduces the unconscious bias that enters any process where decisions are made informally; and it produces data that can be used to improve future hiring decisions. Organisations that have moved to structured, documented assessment are finding that their quality-of-hire metrics improve alongside their compliance posture.
In practice, this means:
- Structured assessment before keyword matching. Design your process so that candidates are evaluated on what they can demonstrate before you weight what they claim. This reverses the traditional sequence.
- Adaptive depth questioning as the primary screen. The question is not whether AI can conduct an assessment. It is whether the assessment is designed to reveal genuine capability rather than rehearsed presentation.
- Consistent criteria across every candidate. A shortlisting decision that cannot be explained with reference to documented evidence is a decision that creates legal and operational exposure. Consistency is both fairer and more defensible.
- Human judgment at the point of decision. AI can structure, score, and document the assessment. The hiring decision, and the accountability for it, remains with your team. This is not just ethically correct. Under the EU AI Act and EEOC enforcement priorities, it is a legal requirement.
- Audit trail from first application to offer. The organisations that will navigate the next wave of regulatory scrutiny are those that can produce a complete, documented record of how every shortlisting decision was made and what evidence supported it.
Sources
- The Economist (2026), reporting significant increases in job application volumes following the rise of generative AI tools. Also: LinkedIn Global Talent Trends Report (2026); Gartner CHRO Priorities Survey (2026).
- Willo Hiring Trends Report 2026, based on data from 2.5 million candidate interviews and employer panels including Toyota and Microsoft. Key findings: 37% of employers rate CVs as a dependable indicator; 41% moving away from CV-first hiring; 68% identify live behavioural interviews as the top talent signal; 70% use structured interviews.
- FBI Internet Crime Complaint Center (IC3), Public Service Announcement PSA250723-4, July 23 2025. North Korean IT Worker Threats to US Businesses. ic3.gov
- US Department of Justice, press release June 30 2025, coordinated enforcement actions targeting North Korean IT worker schemes. Also: Crowell & Moring LLP client alert, September 2025.
- Mandiant / Google Cloud, CTO Charles Carmakal public briefing at annual security conference, May 2025. DOJ prosecution records, 479 documented corporate victims as of mid-2026.
- Gartner (2026), projects 1 in 4 candidate profiles will be fake by 2028. Experian 2026 Future of Fraud Forecast, employment fraud rated second-highest threat category. Also: GetReal Security Enterprise Survey 2026, 41% of enterprises report having hired fraudulent candidates.
- EU AI Act, Regulation (EU) 2024/1689, Annex III, Category 4: Employment and workers management.
- EU AI Act Digital Omnibus Revisions, political agreement reached 7 May 2026. Travers Smith analysis. Crowell & Moring LLP.
- EEOC AI Guidance Removal & Enforcement, National Law Review, March 2026. Holland & Knight, Artificial Intelligence in Hiring: Diverging Federal, State Perspectives, 2025. EEOC Strategic Enforcement Plan FY 2024-2028.
- Mobley v. Workday Inc., N.D. Cal., March 2026 federal court ruling allowing ADEA age discrimination claims to proceed. HR Dive, March 30 2026. Also: Eightfold AI case, HR Dive.
- US State AI & Employment Legislation: Colorado SB 24-205 (effective June 30 2026); California Civil Rights Council AI regulations (2025), four-year data retention requirement; NYC Local Law 144 (operative 2023), annual independent bias audit requirement for automated employment decision tools.
