The fraud with no victim to telephone you
Almost every fraud control an institution owns was built around a third party: someone attacks a customer, or impersonates one, and the customer complains. A victim calls, a dispute is raised, a chargeback is filed, and the institution investigates.
First-party fraud has none of that. The person who applied, who signed, who used the product and who stopped paying is the real account holder. Nobody is impersonated. Nobody complains. The institution's first signal is an arrears report.
What it is. Deception by the customer, in their own name, to obtain or keep money or credit they are not entitled to: lying on an application, taking credit with no intention of repaying, disputing a transaction they made, or operating an account to extract value and disappear. The product worked exactly as designed; the intention behind it was false.
The definitional difficulty, and it is real. Where does optimism end and fraud begin? A borrower who overstates income because they expect a promotion, who genuinely intended to repay and then lost a job, who disputes a charge they half-remember — none of these is obviously a fraudster. Intention is the element that distinguishes fraud from credit risk, and intention is the thing an institution cannot observe directly. This is why the subject is uncomfortable, and why it is dealt with by inference from patterns rather than by proof in any single case.
Why it disappears from view. Four mechanisms, and they compound:
- It is booked as credit loss. The money was lent and not repaid, so the accounting entry is an impairment. Nothing in that entry says "fraud".
- There is no complaint to trigger a case. Fraud investigation capacity follows reported cases, and nobody reports this.
- It is spread across products. A loan, a card, an overdraft and a buy-now-pay-later facility, each owned by a different team with a different loss budget.
- Admitting it is awkward. A high first-party rate implies the lending decision was wrong, which belongs to someone with a growth target.
So the headline fact of this course: a substantial share of what an institution records as credit loss is actually fraud, it is never counted as fraud, and because it is never counted it is never resourced. That is a measurement failure before it is a control failure.
The forms it takes
Application fraud. Misstating income, employment, address, residency status, existing debts or the purpose of borrowing, to obtain a product or a limit that would otherwise be refused. The identity is genuine; the facts around it are not. The most common and the most tolerated, because the applicant is a real person who may well pay.
Never-pay. Credit taken with no intention of a single repayment. Distinctive because the account behaves abnormally from the first day: no payment, often immediate full utilisation, and frequently an application that was unusually complete and unusually fast.
Bust-out. The patient version, and the expensive one. An account is opened and operated impeccably for months or years — small balances, early payments, no missed instalments — so that limits rise and more products are granted. Then every facility is drawn to the maximum in a short window and abandoned. The behaviour that precedes a bust-out is indistinguishable from an excellent customer, which is precisely why limit increases are the mechanism it exploits.
Loan stacking. Multiple applications to multiple lenders within a short window, timed so that each lender's bureau check happens before the others' drawdowns appear. Not necessarily fraud at any single lender; the fraud is in the aggregate, and no single lender can see it.
Friendly fraud and chargeback abuse. Disputing a genuine transaction — claiming non-delivery, non-recognition or non-authorisation for something the customer did and received. Sits awkwardly between fraud and dispute handling, and the incentive for the institution is usually to refund and move on.
Facility abuse. Kiting between accounts, deliberate overdraft running, exploiting a processing window or a promotional term. Small individually and systematic in aggregate.
Synthetic identity, which is the next lesson, because it is the one that defeats the controls built for all of the above.
A note on the honest boundary. Several of these shade into hardship. A customer who draws everything down and stops paying may be busting out, or may have lost their income in the same week. The distinction is in the pattern before the event — the prepared application, the choreographed limit increases, the simultaneous drawdown across lenders — not in the fact of non-payment. An institution that cannot tell the two apart will either miss the fraud or treat distressed customers as criminals, and both are failures.
Loan stacking is worth pausing on, because it is the one form no lender can solve alone. It works on a timing gap: a credit bureau learns about a new facility when the lender reports it, which may be days or weeks after drawdown. Applications made inside that window each see a borrower who does not yet have the other loans.
So the control is not a better model. It is shared data, faster: near-real-time reporting of applications and drawdowns rather than monthly cycles, and consortium databases where lenders contribute confirmed application-fraud cases and query each other's.
Both exist in some markets and not in others, and where they are absent the reason is usually commercial rather than technical — an application is competitive intelligence, and a lender that shares early tells rivals where its customers are. That is a real tension and worth naming honestly, because it explains why a solved problem stays unsolved.
Synthetic identity, as a credit problem
A synthetic identity is a person who does not exist, assembled to look like one who does. It is usually built from a real identifier — a national identity number, a tax number, sometimes one belonging to a child or someone who does not use credit — combined with a name, date of birth and address that do not belong to it. Sometimes it is wholly fabricated.
Why identity verification does not stop it. This is the part that surprises people. The identity is not forged; it is cultivated:
- A thin or absent credit record is created by the first application, which is declined but leaves a footprint.
- Small facilities are obtained — a secured card, a prepaid product, an authorised-user position on someone else's account.
- Those facilities are operated well for months, building a genuine repayment history.
- Within a year the identity has a bureau file, a credit score and a transaction history. It is now more verifiable than many real people, particularly the young, the recently arrived and the under-banked.
By the time a lender meets it, the identity passes every reasonable check — because the checks test whether a record exists and is consistent, and this record does exist and is consistent. Nothing is being impersonated, so there is no victim to notice and dispute.
How it ends. A bust-out. Facilities drawn, nothing repaid, and no one to pursue — the address is accommodation, the employer does not exist, the person cannot be found because the person was never there. The account goes to collections, collections fail, and the balance is written off as credit loss. The institution's fraud statistics never record it.
Where it connects to eKYC. The onboarding and biometric side — deepfakes, injected video, what a camera can be shown — is covered in the identity fraud course linked at the end of this one. The distinction worth holding: eKYC asks is this a real person, present and matching their document; a synthetic identity may well pass that, because the operator is a real person using a cultivated identity that is not theirs. Liveness proves presence, not entitlement.
The scale of the identity problem generally. FinCEN's analysis of 2021 Bank Secrecy Act data found approximately 1.6 million identity-related reports, 42 per cent of around 3.8 million total reports, equivalent to $212 billion in suspicious activity. It separates three ways identity processes are exploited — impersonation, circumvention and compromise — and records that fraud was the most reported typology, with about 1.2 million instances of general fraud among 14 typologies. That report is about identity-related suspicious activity rather than first-party fraud specifically, and it is the clearest public measurement of how much of this reaches the financial system at all.
What actually detects it
Verification of a single application will not. The signal is in behaviour over time and in relationships between applications.
At application:
- Velocity across the institution. The same address, telephone number, device, employer or bank account appearing across multiple applications, especially within a short window.
- Data that is too clean. A perfectly complete application, entered quickly, with no corrections and no hesitation, from someone with a thin file. Real people fumble forms.
- Internal inconsistency over time. Income, employer or address that differs from what the same person told you for a different product last year.
- Identity age against person age. A credit file whose earliest entry is recent while the stated date of birth is not, or the reverse — a file older than the person could plausibly have.
- Contactability. An employer that cannot be reached, an address that is a mail-forwarding service or serviced accommodation, a telephone number with no history.
In behaviour after opening:
- Utilisation immediately at or near the limit, particularly on day one.
- Payment patterns that are too perfect — always the minimum, always on the same day, never varying — on an account whose holder has no other relationship with you. Real financial lives are noisier.
- Limit increases accepted instantly, every time, across products.
- Cash-equivalent spending: cash advances, gift cards, gambling, crypto purchases, high-resale goods.
- A cluster of accounts behaving in step — the strongest signal available, and invisible to any single-account view.
At the point of bust-out, it is already late, but: simultaneous maximum drawdown across facilities, a sudden change of contact details beforehand, and a final flurry of applications elsewhere.
What makes the difference institutionally. Three things, in order of effect:
- A single view across products. Bust-out exploits the fact that the card team, the loan team and the overdraft team each see one facility. One customer view is worth more than any model.
- An identity graph. Linking applications by device, address, telephone, employer and bank account. Fraud rings reuse components because fabricating entirely fresh ones is expensive.
- Feeding confirmed cases back. A first-party case identified in collections must reach the application models, or the same pattern is approved again next quarter.
And the caution that belongs here. Every one of these signals also describes ordinary people: a young person with a thin file, a recent migrant with no address history, a gig worker whose income genuinely varies, someone using a serviced address because they move for work. Used mechanically, this becomes a model that declines the poor and the mobile. The discriminating factor is combination and pattern, never any single flag.
Why it is also an anti-money laundering matter
First-party fraud is usually run entirely inside credit risk and collections. There are two reasons that is wrong, and one of them is a legal obligation.
The proceeds are criminal proceeds. Money obtained by deception is the proceeds of crime, and what happens to it afterwards is money laundering. An institution that identifies a bust-out and books it as a write-off has identified a predicate offence and reported nothing. Where there are reasonable grounds to suspect, the reporting obligation applies — it does not contain an exception for losses the institution bore itself.
The same infrastructure serves other crimes. Synthetic identities and cultivated accounts are not built only for credit. They are used for mule activity, for laundering third-party proceeds and for sanctions evasion. FinCEN's identity analysis records third-party money laundering among its most-reported typologies, alongside general fraud, false records and identity theft. An institution that treats a fabricated identity purely as a lending problem has found a tool and thrown away everything else it was used for.
What to do about it, concretely:
- Classify before you write off. A loss review that asks "was this fraud?" before the impairment is posted is the single structural change. Without it, the question is never asked by anybody.
- Give first-party cases a route to the AML function, and agree in advance who assesses them for reporting. This is the convergence problem in miniature, and the account-closure caution applies: closing and writing off destroys the trail.
- Report what you find, naming the suspicion. "Suspected first-party fraud and bust-out across four facilities, with the identity showing characteristics of cultivation" tells an analyst something; "account written off" tells them nothing because they never see it.
- Count it. Report first-party fraud as a number in its own right, separately from credit loss. Until it is a number, it is nobody's problem and it gets no resource.
- Watch the fairness boundary. Distinguish deliberate deception from hardship, and be able to show how. A programme that cannot tell them apart will hurt people who are already in trouble.
The judgement to take away. First-party fraud is the clearest example in financial crime of a problem that is invisible because of how it is recorded, not because of how it is hidden. The fraudster took no special trouble to conceal anything; the institution's own accounting did the concealing. Changing one question in the loss review — was this fraud? — surfaces more of it than any model will.
And one thing to get right before any of that: what being recorded as a first-party fraudster does to a person.
A fraud marker on a shared database is not an internal note. It can mean refused accounts across the market, refused credit, sometimes refused employment, for years — applied on the strength of an inference about intention that nobody proved, in a case the customer was never told about.
That matters here more than almost anywhere else in financial crime, because the whole subject turns on intention, which cannot be observed. The institution is deciding, from a pattern, that a customer meant to do something. Sometimes that is well evidenced. Sometimes it is an arrears report and a model score.
So the discipline is procedural: a stated evidential threshold for recording a marker, a decision made by someone other than the person who built the case, a written reason, and a route by which the customer can learn of it and contest it. A programme with none of those has the power to end someone's access to banking and no process around using it.
Twenty questions
Answer all twenty questions, then check your answers. You need 18 out of 20 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.
Your answers
Your certificate of completion
Enter your name as you want it to appear, then save the certificate as a PDF. In the print window, choose Save as PDF. A certificate is issued once per completion of the course.
Rate this course
How would you rate this course? Your rating is anonymous — just a number, with no name and no account.
Bring this to your team
Your colleagues can take this course too — it is free, needs no sign-up, and nothing they enter leaves their browser. If your institution wants the same ground covered as structured training for a team, with an instructor and your own cases, we run courses in a classroom or online for groups of ten or more.
Talk to us about training your team → See the paid courses →
Saolix does not record who takes this course, so it cannot verify these certificates. The certificate confirms completion of a free self-paced course and is not an accredited qualification.
Standards and frameworks covered
The published frameworks this course teaches toward. Saolix is not affiliated with, endorsed by or accredited by any of these bodies; naming a standard is a citation, not a claim about its author.
Sources
The official documents this course relies on. Laws and guidance change, so check the current version.
- Financial Trend Analysis: Identity-Related Suspicious Activity — 2021 Threats and Trends (January 2024) · Financial Crimes Enforcement Network (FinCEN), US Treasury
- Payments Fraud Insights: Synthetic Identity Fraud in the U.S. Payment System (July 2019) · Federal Reserve System
- Payments Fraud Insights: Detecting Synthetic Identity Fraud in the U.S. Payment System (October 2019) · Federal Reserve System
- Synthetic Identity Fraud Mitigation Toolkit · Federal Reserve System
- Illicit Financial Flows from Cyber-enabled Fraud (November 2023) · FATF, Egmont Group and INTERPOL
- A Framework for Managing Fraud Risks in Federal Programs (GAO-15-593SP) · US Government Accountability Office
- GovS 013: Counter Fraud (Government Functional Standard) · Cabinet Office (UK)
- 2025 Report on Payment Fraud (December 2025) · European Banking Authority and European Central Bank
- The FATF Recommendations (Recommendations 1, 10 and 20) · Financial Action Task Force



