Deepfake fraud rising as headline mortgage risk
The second quarter of 2026 produced a notable shift in fraud-related mortgage lending risks, according to wire and title fraud prevention firm FundingShield, which monitors mortgage fraud activity across its more than $120 billion portfolio.
At an average of 2.3 issues per problematic transaction, the prevalence of fraudulent activity in the second quarter remained below a record-high average of 3.2 issues clocked in the fourth quarter of last year.
But 46% of transactions with wire and title fraud risks were largely flagged for compliance-related issues around documentation. This trend underscores what the company described as a “meaningful” shift in fraud risks, “moving beyond instruction-level defects into identity and payoff-level fraud,” including the rise of deepfake seller impersonation.
Deepfakes are synthetic visual or auditory media creations that mimic people’s or companies’ appearance or voice in a highly realistic manner, making it more challenging to distinguish fraudulent actors from genuine counterparties in mortgage transactions.
Approximately 45.3% of transactions were flagged for issues in the second quarter, compared to 43.7% in the first quarter. Closing protection letter discrepancies accounted for about 47.5% of defects, compared to about 43.5% the prior quarter.
“A record share of Gen Z, first-time, FHA-heavy borrowers entered the market at the same moment FHA delinquencies climbed, mortgage payoff fraud losses grew, and deep-fake-enabled seller impersonation moved from theoretical to active,” said Ike Suri, CEO of FundingShield, in commentary released alongside the firm’s second-quarter fraud analytics report.
Loans insured by the Federal Housing Administration have been under pressure since 2024 but worsened late last year as pandemic-era loss mitigation programs ended, which has triggered an accelerated flow of FHA loans through default and foreclosure processes.
At the same time, Fannie Mae and Freddie Mac rate-lock volumes fell below 50% market share in June for the third consecutive month, reflecting the growth of so-called “nonagency” loans like non-qualified mortgage (non-QM) products and home equity. Nearly 1 in 5 rate locks was for a non-QM loan in June, surpassing the FHA-lock share.
Suri referenced that shifting production mix in his assessment of evolving fraud risks, including higher relative volumes of investor-purpose loans, which have consistently higher rates of fraud-flagged activity, according to real estate analytics firm Cotality.
Advancements in artificial intelligence have amplified the variety and volume of risks confronted by counterparties across loan closing workflows, contributing to an increase in cybersecurity insurance premiums, FundingShield noted. “Identity and payoff-fraud flags” were described as a “new and growing category” of mortgage fraud risk.
Flagged wire fraud risks declined over the quarter, however, from appearing in 6.92% of transactions to 6.45% during the second quarter.
“The broader threat environment is not easing,” said Suri. “AI-enabled bad actors are widening uncertainty across mortgages, securities and trading,” he added, while private credit markets are “racing toward daily pricing and secondary liquidity.”
