AI is making mortgage fraud harder to spot, Equifax warns

Canada’s reported mortgage application fraud rate is falling, but increasingly sophisticated fake documents may be allowing some fraudulent applications to escape detection, warns Equifax Canada’s Carl Davies.

In an interview with Canadian Mortgage Trends, Equifax’s head of fraud and identity said AI is making falsified income and employment documents harder to spot, raising questions about whether the decline reflects the full extent of mortgage fraud.

As CMT previously reported, Equifax’s mortgage application fraud rate fell to 0.20% in Q2 2026, with Ontario recording the highest provincial rate at 0.28%, followed by Alberta at 0.19%.

Davies says even those small percentages represent “hundreds of millions of dollars” in mortgage applications.

“Those numbers, based upon the application volume, get very high very quickly,” he says. “While it might look like a small percentage, the impact on lenders is very, very high.”

Lower reported fraud may not tell the full story

Davies suspects some fraud that escapes detection during underwriting is later being classified as credit delinquency when borrowers fall behind on payments.

“A good portion of what we’re considering as credit delinquency — where people traditionally have no malintent but are unable to pay — we think a fairly significant amount of that now is fraud,” he says.

“Because of the sophistication of fraud at the point of adjudication, it’s actually getting through onto the books and sitting there, so I do think in certain areas we’re underreporting those fraud rates.”

Most mortgage fraud involves applicants misrepresenting their finances

Unlike credit card application fraud, which Davies says primarily involves third parties using someone else’s identity, mortgage fraud is dominated by applicants misrepresenting their own information.

“The vast majority — we’re talking 80, 90% — of the [mortgage and auto loan] fraud that we see is first-party based,” Davies says. “It’s people either looking for shelter or wanting to get the vehicle that they want, but probably shouldn’t be able to afford, and falsifying some of that information in order to make themselves look like a better credit risk.”

Davies says common forms of mortgage fraud include falsified employment letters and income statements, as well as undisclosed debts.

Applicants may invent a second or third job paid primarily in tips or cash, or alter income statements to inflate their earnings, Davies says, while AI makes it easier to produce convincing documents to support those false claims.

“Typically, that manifests itself as falsified documentation, which if you go back 10 or 15 years, wasn’t super easy to do,” Davies says. “With the advent of AI models, the ability to create evidence documents that are almost imperceptible to humans is actually very easy.”

Brokers have a role in catching suspect applications

Davies says lenders detect most mortgage application fraud, but brokers can help identify suspicious files before submission.

He says increasingly sophisticated and affordable fraud-detection tools, including those using AI, can help brokers flag suspect documents before spending more time on an application.

“When we’re doing some of these fraud investigations with lenders or brokers or auto dealers, it’s funny because in many cases they’ll say, ‘you know something kind of didn’t feel right,’” Davies says. “I think there’s absolutely ways that we can empower brokers to say, ‘hey, I’m going to challenge this one because it looks a little problematic.’”

“Fraud is not something that’s going to go away; it’s been around forever and probably will be around forever,” he says. “The more people can work together and trust their instincts, the harder we’re going to make it for both organized crime and for those first-party fraud actors.”

Visited 157 times, 157 visit(s) today

Carla Davies equifax equifax Canada fraud mortgage fraud

Last modified: October 13, 2026