Fake candidates: when the person you interviewed is not the person you hired
Remote hiring made it possible to interview one person and employ another. How candidate fraud works, and how to verify without profiling applicants.

Hiring used to include a physical event. The candidate came to a building, was seen by a receptionist, shook several hands, and later collected a laptop from somebody who looked at their face. None of those steps were designed as identity verification, and all of them worked as identity verification.
Remove them and a gap opens that most hiring processes have not closed. An entire recruitment cycle can now run from first application to first payroll without any one person confirming that the human on the screen is the human named on the contract, that the same human sat each stage, or that the same human will be doing the work.
The existing library on this site covers the risk facing applicants, in fake job offers. This article is the inverse: the organization is the target, the application is the attack, and it is best understood as a social engineering problem rather than a vetting one.
Four different frauds that look the same
Lumping these together is the usual mistake, because they need different detection and carry very different consequences.
The proxy interview
One person interviews, another works. This is the oldest form and it remains the most common, particularly for technical roles where the assessment is a skills test that can be sat by anyone. Sometimes the substitution is complete and a different person appears on day one. More often the hired person is real and genuinely wants the job, with an unseen helper who passed the test and may continue to do parts of the work.
What gives it away is almost never the interview. It is the discontinuity afterwards: capability on the job that bears no relation to the assessment score, reluctance to be on camera in team meetings after being comfortable at interview, or work that arrives in a style inconsistent with how the person speaks about it.
Identity substitution
Here the name, documents and history belong to someone else, either a real person whose identity has been stolen or a fabricated composite. The motive is usually that the actual worker cannot pass a check: a previous dismissal, a regulatory bar, immigration status, or a criminal record relevant to the role.
This one has a second victim. The person whose identity was used may find employment and tax records they cannot explain, which is the problem described from their side in employment identity fraud. If you discover it, that person needs to be told.
The intermediary placement
An agency or outsourcing operation presents a candidate under a false name, with an invented history, or presents one person's work as another's. The motive is the placement fee, and the pattern is visible in volume rather than in any single case: several candidates from the same source with similarly structured histories, identical formatting, overlapping references and the same contact number appearing against different names.
Infiltration for access
The most serious category, and the one that changes how you respond. Here the employment itself is the attack. The objective is a legitimate account, a company laptop, access to source code or customer data, and a salary paid to an account the organization has verified itself. The FBI has published repeated advisories about state-linked schemes in which remote IT workers obtain employment under false identities, with motives including revenue generation and, in documented cases, extortion after copying company data.
Those advisories make one point that is worth carrying over to every category here: no single indicator reliably identifies a fraudulent remote worker, and much of the identity information presented comes from stolen real identities. Any approach that depends on spotting one tell will fail.
Where processes actually break
The weaknesses are mundane and nearly universal.
- Nobody checks continuity between stages. A screening call, three interviews and an onboarding session may involve four different interviewers and no record of whether the same person attended each one.
- Identity verification happens once, with nobody comparing it to anything. A document is uploaded during onboarding, filed, and never matched against the person who interviewed.
- References are verified by email. An email reference from a domain the candidate supplied is a document the candidate controls.
- Equipment goes to an unverified address. A laptop with standing access to the corporate network is couriered to an address nobody has checked, which in several documented schemes was a forwarding service or a paid local facilitator.
- Payroll details are set after hire with no verification. A bank account change in the first week is routine administration to HR and is also the point at which a placement fraud monetizes, which is the overlap with payroll diversion fraud.
- Recruiters are a phishing surface in their own right. The team opening attachments from strangers all day is the team an attacker targets, as set out in phishing aimed at HR and recruiters. A compromised recruiter mailbox can insert a candidate into a process directly.
Verification that is proportionate and consistent
The design constraint matters as much as the controls. Checks must be applied identically to every candidate for a role, documented, and based on process rather than on judgments about how someone sounds, where they appear to be, or what they look like. Screening built on those judgments creates discrimination exposure and does not work, since the identities presented are frequently real ones belonging to people who would pass any such instinct.
- Verify identity with a liveness check, at a point you control. A government document matched against a live capture, through a recognized verification provider, rather than a photograph of a document emailed in.
- Establish continuity across the process. Record who attended each stage and confirm at onboarding that it is the same person. In practice this means one short video session at the end, with the verified identity document to hand, conducted by someone who attended an earlier stage.
- Verify employment history through the employer, not the candidate's contact. Call the organization's published switchboard. If a listed former employer has no verifiable existence, that is a finding in itself, and the same public-register checks you would run on a new supplier apply to a CV.
- Make the technical assessment conversational. Ask candidates to explain decisions in their submitted work, live, with follow-up questions. A proxy who wrote the solution can do this and a substitute cannot, and it is a better interview anyway.
- Verify the delivery address before shipping equipment. Treat the first equipment dispatch as a control point, and reconcile the address against the verified identity and the payroll record.
- Freeze payroll detail changes for an initial period. Or at minimum require verification for any change in the first month, with notification to the employee through a separate channel.
- Watch the first thirty days as a security window. Unusual bulk access to repositories or customer records, use of personal cloud storage, or remote access tooling appearing on a new starter's device are all worth an alert. This is where least privilege earns its cost, because it bounds what a fraudulent hire can reach before anyone notices.
Candidate email or document that looks off? Paste the message or link into CyberWatch AI Scan for a free check before you progress the application.
Signals that are worth something, and ones that are not
Useful signals are procedural: refusal to appear on camera at any stage, audio consistently out of step with video, a different voice or appearance between stages, an address that resolves to a mail forwarding service, references reachable only by email, a payroll account changed immediately after hire, and a request to use a personal device instead of company equipment.
Signals that are not worth anything, and are actively harmful to rely on, include accent, name, apparent location, a slightly unusual CV format, imperfect written English, and a camera that fails once on a bad connection. Building screening around those produces false accusations against legitimate applicants and misses the schemes that matter, because the people running them present documents belonging to real people with unremarkable profiles.
When a hire turns out to be fraudulent
Handle the security question before the employment question, because the sequence is easy to get wrong.
Revoke access and recover equipment, but preserve the account's activity before deleting it: authentication logs, data access and transfer records, repository activity and email. Establish what was reached and whether anything left the organization, which determines whether this is a hiring failure or a data breach with notification obligations.
Then take the employment and legal steps with advice, since dismissal, pay already made, and any sanctions exposure where a state-linked scheme is suspected all need handling properly. Notify the person whose identity was used if one has been identified. Report through the routes in reporting a cyber incident, because these schemes recycle identities and documents across many employers, and your report is often what links them.
Finally, fix the specific gap rather than adding a general warning to the hiring handbook. In nearly every case the answer is one of three things: nobody confirmed the same person attended each stage, nobody verified an address, or nobody re-verified a bank account change.
Frequently asked questions
What is a fake candidate?
An applicant whose identity, or whose ability to do the work, is not what the employer was shown. It covers proxy interviewing, identity substitution, fabricated work histories and candidates presented under a false name by an intermediary.
What is proxy interviewing?
One person sits the interview while another does the job. A skilled interviewee passes the technical assessment and a different, less capable person turns up on day one, or the capable person stays on as an unseen helper.
Are deepfakes really being used in interviews?
Live video manipulation is being attempted, and the FBI has warned that generative AI is lowering the effort and cost of creating convincing fraudulent material for financial crime. In practice the weaker signals are still behavioral and procedural rather than visual.
Why would anyone bother faking a candidate?
Several motives: a person who cannot pass vetting wants the salary, an outsourcing intermediary wants the placement fee, an individual wants to hold several full-time jobs at once, or an organized group wants paid access to a company's systems and data.
What is the single most useful check?
Verifying identity against a government document with a liveness check at an independent point in the process, then confirming the same person appears at onboarding and at first-day equipment collection.
How do we avoid discriminating against legitimate applicants?
Apply the same checks to every candidate for the role, base them on documented process rather than on how someone sounds or looks, and never build screening around nationality, accent or location. Inconsistent checks create legal exposure and catch nobody.
What are the real red flags?
Refusal to be on camera, audio that lags the video, a different person at onboarding, a changed bank account immediately after hire, an address that is a forwarding service, and references that only answer by email.
What should happen if a hire turns out to be fraudulent?
Treat it as a security incident as well as an employment matter. Revoke access first, preserve the account's activity logs, and establish what was accessed before deciding how to handle the termination.
Sources
- Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud, FBI Internet Crime Complaint Center
- North Korean IT Worker Threats to U.S. Businesses, FBI Internet Crime Complaint Center
- North Korean IT Workers Conducting Data Extortion, FBI Internet Crime Complaint Center
- Internet Crime Report 2024, FBI Internet Crime Complaint Center
- Report to the Nations: occupational fraud research, Association of Certified Fraud Examiners
- Impersonation (T1656), MITRE ATT&CK
- Establish Accounts (T1585), MITRE ATT&CK
- 10 Steps to Cyber Security, UK National Cyber Security Centre


