← Back to Content

Synthetic Identity Fraud: Detecting Fabricated Users

Synthetic identity fraud uses fabricated identities that pass credit checks. Learn the red flags, why KYC misses them, and how browser signals help.

Robin
fraudsynthetic identity fraudsynthetic identity theftfraud detectiondigital identitybrowser fingerprinting
Synthetic Identity Fraud: Detecting Fabricated Users

Synthetic identity fraud is the one common kind of fraud where nobody calls to report it. There is no victim to dispute the account, nothing is flagged as stolen, and the applicant passes the identity checks you run, because the identity was built to pass them. The loss arrives months later, all at once, on an account that looked like one of your better customers.

Below: what a synthetic identity is, how one is grown to bust-out, the red flags in your own application data, and why identity verification alone leaves a gap.

What Is Synthetic Identity Fraud?

Synthetic identity fraud is the use of a fabricated identity, assembled from invented details and real but unrelated ones, to open accounts that belong to no actual person. The fraudster builds the identity's credibility over time, then uses it to borrow or buy without ever repaying.

The Federal Reserve's industry-recommended definition puts it more formally: "the use of a combination of personally identifiable information (PII) to fabricate a person or entity in order to commit a dishonest act for personal or financial gain."

A synthetic identity usually takes one of three forms:

  • Fully fabricated: every element is invented, including the name, date of birth and Social Security number (SSN).

  • Compiled, or "Frankenstein": a real SSN, often one belonging to a child, an elderly person or someone deceased, combined with an invented name and date of birth. The SSN's real holder is unlikely to check a credit report any time soon.

  • Manipulated: a real person's own details, slightly altered (a changed digit in the SSN, a different date of birth) to create a fresh credit file and hide a poor history.

Synthetic identity theft vs. identity theft

Synthetic identity theft is the same practice under another name, though "theft" fits poorly. Traditional identity theft uses a real person's identity without permission, to open accounts in their name or, in account takeover, to get into accounts they already hold. Sooner or later that person notices and complains.

A synthetic identity has no owner to notice. Even a borrowed SSN often belongs to a child who will not see a credit report for years, so the feedback loop that exposes ordinary identity theft never starts.

Why a Synthetic Identity Passes the Checks Built to Stop It

Identity verification against credit bureau and public records asks one question: does this identity have a consistent history? Against stolen identities that question works well. The details are real, so mismatches appear (a new address, an unfamiliar phone number), and the true owner eventually disputes the account.

A synthetic identity answers the question by acquiring a history. Its first credit applications are usually declined, but an application on an identity a bureau has never seen can be enough to start a file for it. The fraudster applies again, gets a small line, and uses it normally. After a year or so the identity has a consistent name, date of birth, SSN, address and payment record, which is exactly what verification looks for.

Two older checks have also weakened. Since June 25, 2011, the Social Security Administration (SSA) has assigned numbers randomly, so a recent SSN no longer reveals where or roughly when it was issued. The SSA's consent-based verification service lets lenders confirm that a name, date of birth and SSN match its records, which narrows the "Frankenstein" gap. But a match does not prove the applicant is the holder, and access is limited to financial institutions and their service providers, which rules out most marketplaces and subscription businesses.

None of this is an argument against identity verification. It is the right control for stolen identities, and it catches crude synthetic ones. But it answers "does this identity look real?", and a well-made synthetic identity is designed to make the answer yes.

The Synthetic Identity Lifecycle, From Application to Bust-Out

Synthetic fraud follows a recognisable pattern. Timescales vary, so treat these stages as a typical shape rather than a schedule.

  1. Creation (day one). The operator assembles a name, date of birth, SSN, address, phone number and email address. The address is often a mail drop or a property shared with other synthetic identities.

  2. First foothold (months one to three). Early applications are declined, but they seed a credit file. The identity gets its first small line: a secured card, a retail store card, or an authorised-user slot on someone else's established card ("piggybacking"), which lends it that account's history.

  3. Good behaviour (roughly months three to eighteen). Small balances, paid on time, every month. The score climbs and the file looks like a careful young borrower's.

  4. Growth and stacking. Limit increases follow, then new accounts at other lenders, buy now, pay later (BNPL) providers and marketplaces. One identity may hold a dozen accounts, each individually in good standing.

  5. Bust-out. Every line is drawn down at once: resalable goods, cash advances, sometimes a bad payment that briefly frees up the limit to be spent again. Then the identity goes silent.

The loss lands at the transaction stage, where payment fraud detection operates, but the decision that mattered was made a year earlier. Because nobody disputes anything, the loss is often booked as an ordinary credit write-off, so the problem stays under-counted.

The aggregate is still large. TransUnion estimated US lender exposure to suspected synthetic identities at $3.3 billion at the end of 2024, across auto loans, credit cards, retail cards and unsecured personal loans, an all-time high in its data.

For detection, the economics matter more than the total. One synthetic identity takes a year or more to yield a few thousand dollars, so the work only pays at volume: a profitable operation runs dozens of identities in parallel, and that is its weakness.

Red Flags of Synthetic Identity Fraud

Red flags fall into two groups. The first is what a bureau or verification provider can see. The second is what only your own systems can see, and it is the group most guides leave out.

Identity-level red flags

  • Thin or new credit file on an adult: a 38-year-old applicant whose file is less than two years old.

  • SSN inconsistent with date of birth: a number issued before the applicant was born. This only applies to numbers issued before 2011.

  • Shared contact details: the same address, phone number or email pattern appearing across unrelated identities.

  • Unverifiable employment or income: employers that cannot be found and income that cannot be corroborated.

  • Authorised-user stacking: a file built mostly from authorised-user lines on strangers' accounts.

Behavioural and environmental red flags

  • Many applications from one browser environment: several "different" applicants arriving from the same browser.

  • Identical browser and device characteristics: matching configurations across applicants with no plausible connection.

  • Unnaturally fast or uniform form completion: every field filled at the same speed, no corrections, no hesitation on questions a first-time applicant would pause over.

  • Clustered signups: bursts of applications in a short window, often at odd hours for the stated location.

The second group complements the first rather than replacing it. An identity can be flawless on paper and still arrive from the same browser as thirty others.

Synthetic Identity Fraud Detection: Looking for the Operator, Not the Identity

Most detection tooling scores one application at a time, which suits stolen identities and fails against synthetic ones, because each is individually convincing. The reliable artefact is not a flaw in any single identity but the repeated environment they are operated from.

Correlation, not fabrication

To see that environment, you need an identifier for each application that does not depend on what the applicant typed. Browser fingerprinting provides one: it combines characteristics of the browser and the device it runs on into an identifier that stays the same across visits in that browser, even after cookies are cleared. Our browser fingerprinting guide explains how it works.

Store that identifier against every application and a new question becomes answerable: how many applicants share a browser environment? A genuine applicant pool rarely puts forty identities through one browser. The signal catches correlation, not fabrication. It cannot tell you that any one identity is fake, only that many were operated together, which is the thing bureau data structurally cannot see.

Some operators automate their submissions, and bot detection will catch part of that traffic. Many synthetic applications are submitted by hand, carefully, precisely because the operator is investing in each identity. Bot signals alone will miss them.

Where ThumbmarkJS fits

ThumbmarkJS supplies this identifier and is used on 60,000+ websites. Your signup or application page requests it, and webhooks deliver it to your backend for storage with the application record, out of the applicant's reach.

One configuration point matters here. By default the fingerprint includes a coarse network identifier, and synthetic operators often rotate IP addresses, so configure it to ignore the network; otherwise one browser on a new connection looks like a new visitor. The Visitor ID adds continuity, linking fingerprints from the same browser while its local storage survives.

The limits are worth stating plainly. Recognition holds within the same browser on the same device. An operator who switches browsers, moves between machines or uses an anti-detect browser to alter the browser's characteristics appears as someone new, so this raises the cost of each identity rather than ending the operation. ThumbmarkJS does not verify identity, check PII or perform behavioural analysis, and it does not replace know your customer (KYC) checks. It supplies the correlation signal; the decision logic is yours.

How to Prevent Synthetic Identity Fraud

Preventing synthetic identity fraud means layering controls that each catch what the others miss. Ordered by what a business outside traditional lending can realistically do:

  • Verify identity where you can. Bureau and public-record checks, and SSA verification if you are eligible, stop stolen identities and crude synthetics.

  • Correlate by browser environment from the first touch. Record a browser identifier, such as a ThumbmarkJS fingerprint, at signup rather than only at the credit decision, so the history exists when you need to query it.

  • Apply velocity limits across shared attributes. Cap how many accounts can share an address, phone number, bank account or browser identifier within a period.

  • Re-check at moments of growth. A synthetic identity passes onboarding by design. Limit increases, payout changes and unusually large orders are where the bust-out is prepared.

  • Route to review rather than block. A shared browser has innocent explanations: a family computer, a library terminal, a household applying together. Two applicants sharing a browser is normal; fifteen deserves a human look.

No layer is sufficient alone. If your problem is many throwaway accounts rather than a few carefully built ones, that is a different pattern, covered in our guide to new account fraud.

Where Synthetic Identity Fraud Shows Up Outside Lending

Almost every guide assumes the reader issues credit. The same pattern appears anywhere a business extends trust, goods or money before it has been paid in full:

  • BNPL: credit decisions at checkout with light checks, so a synthetic identity can build a record on small plans before a large one.

  • Marketplaces and seller onboarding: fabricated sellers build ratings on small sales, then take payment for goods that never ship. It is the same multi-accounting pattern seen in loyalty fraud.

  • Subscriptions billed in arrears: accounts that pay small invoices on time, then run up a large balance and disappear.

  • Insurance: policies taken out on fabricated people, followed by claims.

  • Rentals: leases signed with fabricated identities, an issue the National Apartment Association addressed in a 2022 white paper on synthetic fraud's cost to rental owners and operators.

Conclusion

A synthetic identity is built to survive identity verification, so checking the identity harder has diminishing returns. The durable signal is the environment the identities are operated from: fabrication only pays at volume, and volume leaves a trail in your own application data.

In practice, adoption looks like this:

  • Who installs it: the engineer who owns the signup or application flow adds the fingerprint request and stores the identifier with each application record.

  • First useful output: a query over the first few weeks of applications showing how many share a browser environment. The integration takes days, not a quarter.

  • What happens next: the fraud or risk owner sets a review threshold and decides what a shared environment triggers.

Our use cases overview shows how other teams apply the same signal, and the free API tier lets you start collecting the signal before choosing a plan.

Frequently Asked Questions

What is the difference between synthetic identity fraud and identity theft?

Identity theft misuses a real person's identity, and that person eventually disputes the activity. Synthetic identity fraud uses an identity that belongs to nobody, so there is usually no victim to raise the alarm.

Can identity verification detect a synthetic identity?

Sometimes, and less often than you would hope. Verification catches crude synthetics, but one used carefully for a year or more has exactly the consistent history verification looks for.

What is a bust-out scam?

The final stage of synthetic identity fraud: after months of on-time payments have raised its limits, the identity draws down every line at once and disappears. With no real person to pursue, the balance is written off.

How do you detect synthetic identity fraud at signup rather than after the loss?

Look at what links applications together. Recording a browser identifier and checking shared addresses, phone numbers and timing at signup reveals one operator behind many applicants, long before any account busts out.