Auto lending fraud exposure hit a record $10.4 billion in 2025, up from $9.2 billion the year before, according to Point Predictive’s 2026 Auto Lending Fraud Trends Report. That is nearly five times the level measured in 2010. And here is the part that should worry every lender: most of that money never gets recorded as fraud. It gets booked as a default.
At VerifyPDF, we spend our days looking at the documents behind these applications: the payslips, the bank statements, the employment letters. And in our experience, car finance is where bad documents do the most damage with the least resistance. The channel is built for speed. A buyer sits in a dealership, wants to drive away today, and the lender has minutes (not days) to say yes. That pressure is exactly what fraudsters are counting on.
So why does so much auto loan fraud disappear into default statistics instead of fraud reports? Let’s get into it.
Why fake payslips slip through car finance approvals
Car finance is a fast-decision, high-volume channel. Dealers need funded contracts, buyers expect instant approvals and lenders process stipulations (proof of income, proof of address, bank statements) under real time pressure to win the deal. A manipulated payslip or altered bank statement only has to survive a quick human glance before the vehicle leaves the lot.
That is a different game from mortgage underwriting, where a file might sit with an analyst for days. In auto, the document gets uploaded, eyeballed and approved in the time it takes to fill out the paperwork. Nobody is running document forensics on a Tuesday afternoon when there are six more deals waiting.
And the documents themselves are embarrassingly easy to fake. A payslip is just a PDF. Open it in an editor, change the salary, keep the embedded fonts and layout intact. Now you have a forgery no human reviewer will catch by sight. We wrote about exactly this problem in our analysis of the rising threat of fake bank statements, and it applies doubly to car finance because the review window is so short.
Here’s the uncomfortable math. TransUnion found that auto loan fraud losses run roughly 21 times higher than credit card fraud losses, with the average loss on a fraudulent auto loan running just under $20,000. One bad application costs more than dozens of skimmed cards. A reviewer who waves through one forged payslip a week is quietly bleeding the portfolio.
The fraud patterns specific to vehicle finance
Auto fraud is not one thing. It is a handful of distinct patterns, and they leave different fingerprints in the document file. If you only know one of them, you will miss the rest.
-
Income and employment misrepresentation. This is the big one. Point Predictive’s data puts income and employment misrepresentation at 45% of total fraud exposure, and it grew 21% year over year. A borrower inflates the salary on a payslip, invents an employer or omits existing debt to fix their debt-to-income ratio. Most of these are real people using real names. That is what makes it so hard. The deception is in the document, not the identity.
-
Synthetic identities. Fraudsters blend real data (often a stolen or unused Social Security number, or in Europe a misused national ID) with fabricated names and addresses to build a borrower who looks prime: clean credit file, steady employment, no negative marks. These profiles are patient. They mature across systems, pass traditional checks and fail later, frequently across several lenders at once. Synthetic identity fraud has been one of the fastest rising categories in auto, supercharged by generative AI tools that mass produce supporting documents.
-
Straw borrowers. Someone with clean credit fronts a loan for the real beneficiary, who has no intention of paying. Straw borrower loans often default without a single payment being made, and where a dealer is running the scheme systematically, early payment default rates can exceed 5% of loan production. The credit score is genuine. The intent to repay is fiction.
-
Dealer-driven schemes. Some dealers inflate vehicle values, attach fake employment to borrowers and push deals through in volume. dotData found that systematic dealer risk of this kind can increase a lender’s default risk for those dealers by as much as 500%. The fraud is not in one application. It is in the pattern across a dealer’s whole book.
The common thread? In car finance, the lie lives in a PDF. A payslip, a bank statement, a proof of address. If you cannot verify the document, you cannot catch the pattern.
First-payment default is fraud wearing a costume
Here is the idea that reframes the whole problem. A lot of what gets logged as a default is actually fraud that was never recognized as fraud.
Traditional credit models answer one question: can this borrower pay? They were never built to answer a second question: did this borrower ever intend to pay? Those are not the same thing. A legitimate borrower under stress tends to miss payments later, with patches of recovery and relapse. A first-party fraudster defaults fast and never comes back.
That early, clean break has a name: first-payment default, or early payment default. And it correlates strongly with fraud at application. Point Predictive’s Early Payment Default Risk Index now sits at more than double its 2017 baseline, and more than 70% of these early defaults show evidence of fraud in the original application. In other words, the forged payslip and the missed first payment are usually the same story, told at two different moments.
This is why auto loan fraud is so badly undercounted. The loan funds, the borrower vanishes or stops paying almost immediately, the account rolls to collections and the loss gets filed under credit risk. The fake document that started it all never gets a second look. The portfolio sees a default. It never sees the fraud.
And the timing matters. Subprime auto delinquencies hit record levels in 2024 and 2025, and repossessions jumped an estimated 43% from 2022 to 2024, according to the Consumer Federation of America’s Driven to Default report. When the economy softens, genuine hardship and deliberate fraud both rise, and they get harder to tell apart. The only reliable way to separate them is to look at whether the application documents were real in the first place.
European motor finance faces the same fake-document fraud
If you work in European motor finance and think this is an American story, look closer to home. UK fraud prevention service Cifas recorded more than 444,000 cases to its National Fraud Database in 2025, the highest ever, with identity fraud and facility takeover behind nearly three quarters of them, per its Fraudscape 2026 report. Insurance identity fraud alone rose 26% year on year, and the same synthetic identities and fabricated profiles that hit insurers feed straight into vehicle finance applications.
The mechanics travel well. A fabricated payslip works the same in Manchester as it does in Miami. A synthetic identity built to season across lenders does not care about borders. What changes is the document format, and that is its own trap: a forged UK payslip looks nothing like a forged German one. We process documents from over 90 countries at VerifyPDF, and the variety in payroll layouts, bank statement structures and security features is enormous. No dealer’s finance desk can know what a legitimate document looks like in every format an applicant might hand over.
Why manual document checks cannot keep up
Manual review fails in auto finance for a specific, structural reason: the channel does not give you time, and the documents do not give your eyes anything to work with.
A trained fraud analyst, given a quiet hour, might catch 60-70% of sophisticated forgeries. A finance manager at a dealership, approving a deal while the customer waits, catches far fewer. That is not a knock on anyone’s competence. It is what happens when you ask a human to spot byte-level manipulation in a PDF under time pressure, at volume, all day. Fatigue sets in, error rates climb and the good fakes (the ones that preserve fonts, metadata and layout) sail right through.
We put numbers on this in our comparison of AI fraud detection against manual checks. The short version: humans are good at judgment and bad at scale, and auto finance is all scale. The manipulations that matter (a producer field showing a photo editor instead of payroll software, a creation date that does not match the stated pay period, a balance that does not reconcile) are invisible to the naked eye and obvious to document forensics.
There is also a coordination problem manual review cannot solve at all. When a fraudster fires the same application at five lenders the same day, tweaking the stated income each time, no single reviewer sees the pattern. Each application looks fine in isolation. The fraud only shows up across institutions, and a human flipping through one file will never see it.
How fast document forensics fits a real-time approval flow
The good news: catching this does not mean slowing down the deal. The whole point of automated document forensics is that it runs in the gap you already have.
When an applicant uploads a payslip or bank statement, the document goes through verification before a human ever touches it. The check runs in parallel with the rest of your decisioning, not as an extra step bolted on the end. We covered the mechanics of this in how to verify income documents without slowing onboarding, and the principle is the same in a dealership: the verification finishes before the human is ready to look anyway.
Here is what that looks like in practice for an auto lender:
-
Verify on upload. Every payslip, bank statement and proof of address runs through automated forensics the moment it enters the pipeline. Think of it as a virus scan on an attachment. It is happening whether or not anyone is watching.
-
Score, do not just pass or fail. The document comes back with a risk rating, not a binary yes or no. A clean document funds at full speed. A flagged one routes to a human who now knows exactly where to look, instead of squinting at every file hoping to get lucky.
-
Cross-reference the documents. A payslip claiming a certain monthly salary should match a bank statement showing roughly that amount landing every month. When they do not line up, that is a red flag the documents themselves hand you, if you actually read them at the byte level.
-
Refuse screenshots and scans. Original digital PDFs carry the metadata and structure that make forensic analysis possible. A photo of a screen carries nothing. If an applicant cannot produce an original PDF of a bank statement, that should be a flag on its own.
VerifyPDF checks a document in under 5 seconds and returns a risk rating from “Trusted” to “High risk.” For a high-volume auto lender, that fits inside the approval window without adding friction the borrower can feel. The honest borrower drives away on time. The forged payslip gets caught before it becomes a first-payment default. That is the difference between a portfolio that books fraud as fraud and one that keeps mistaking it for credit risk.
We see the same playbook everywhere documents move fast and money moves faster, from the Boston mortgage fraud ring that used fake payslips and rented tradelines to the broader categories we break down in our guide to the three types of fraud, including first-party fraud. Auto finance is not special in the techniques used against it. It is special in how little time it gives you to catch them.
Stop booking fraud as defaults
Every fake payslip that funds a car loan is a loss waiting to happen, and right now most of those losses are hiding in your default numbers wearing a credit-risk disguise. The $10.4 billion figure is not really a fraud number. It is the size of the blind spot.
If your auto finance approvals run on speed (and they do, because that is the business) then the only way to verify documents without killing the deal is to verify them automatically, in parallel, before a human ever sees them. Try the free document check to see what your applicants’ payslips actually look like under the hood, or get in touch to talk about fitting verification into your approval flow. Because the question was never whether fraudsters are sending you fake documents. It is whether you will recognize the next default for what it really was.