First-Party Fraud Likely Drives Most Identity Fraud Losses

Across recent retrostudies with more than a dozen companies, SentiLink found that more than any other type of identity fraud, first-party fraud is making it onto institutions' books.

• 2.89% — average percent of funded loans/booked accounts that score as high-risk for first-party fraud.
• 5x — on average, institutions likely have 5x more first-party fraud on their books than identity theft or synthetic fraud combined.
• 72% — on average, first-party fraud accounted for 72% of preventable fraud losses across all three types of identity fraud.

So why is so much first-party fraud risk making it past account-opening fraud checks?

Why first-party fraud is hard to define and measure

First-party fraud gets through because it's trickier to detect.

With identity theft and synthetic fraud, there are clear risk signals in the application identity's core PII. And in both cases, it's often relatively easy to confirm — victims report identity theft, and eCBSV can often provide certainty about whether an identity is synthetic.

But first-party fraud is much less clear, especially at the time of application:

• The core identity information is real.
• Even when accounts charge off, it's difficult to confirm whether a given account is first-party fraud or credit loss.
• The strongest risk signals often sit outside of the institution's view, on previous applications to other institutions and in other industries.

One problem is that there's no standardized definition of first-party fraud. Most institutions define it as relating to a borrower's intent to pay back a loan (for example). But intent is impossible to measure. This type of definition leads inevitably to inconsistent labels and continued difficulty flagging first-party fraud risk.

SentiLink defines first-party fraud as the act of presenting a financial institution with accurate core personal information but inaccurate, misleading, or incomplete financial identity information. This definition underpins the way our First Party Fraud Score works — we look for empirical signals of fraud risk associated with the application identity and other application attributes across applications within our network of 500+ partners. And at the institutions where we've been testing this score, we've found that an average of 72% of the fraud losses on their books were associated with high first-party fraud risk (according to our score) at the time of application.

First-party fraud is everywhere

First-party fraud rates were at a flat 2% in the first half of 2026 — quite a bit lower than identity theft rates, but higher than synthetic fraud rates.

But first-party fraud applications are not distributed evenly across industries. Auto lenders and telecom companies in particular face first-party fraud rates much higher than the overall average:

(Wondering what's up with the high first-party fraud rate in auto lending? We have a new whitepaper focused specifically on this problem in the auto lending space).

How to stop first-party fraud

Defeating first-party fraud requires a clear definition that points toward empirical risk signals, and a wide, cross-industry view into identity and fraud that allows you to identify those signals at the point of application. Learn more about how SentiLink defines and detects first-party fraud in another new whitepaper: Real Identities, Real Losses: Reducing First-Party Fraud.

You can also learn more about our observations of first-party fraud over the first half of 2026 in our 1H 2026 Fraud Report.