Financial fraud today is no longer chaotic. It is being optimized by artificial intelligence.
Across the lending industry, a new category of fraud is emerging that combines deepfake video, cloned voices, synthetic identity creation, fabricated employment histories and AI-generated financial behavior into a single engineered persona. These synthetic borrowers are not merely fake identities in the traditional sense. They are algorithmically optimized consumers designed to survive onboarding checks, satisfy underwriting models and disappear once loans are funded.
For banks, credit unions and FinTech lenders, the implications are reshaping the traditional assumptions underlying underwriting, namely that fraudsters will eventually reveal themselves through inconsistency. Digital finance was built around the assumption that more data creates more certainty. Synthetic borrowers invert that premise.
The risk today is that, in an AI-saturated environment, more data may simply create more convincing illusions.
See also: The End of the Artisanal Hack: How AI Industrialized Cybercrime
Bank Fraud Evolves From Theft to Simulation
Synthetic identity fraud is not new. Criminal networks have long combined real and fabricated information, such as valid Social Security numbers paired with fake names, to create identities capable of building credit profiles over time. What has changed is the scale, speed and realism AI now enables.
A synthetic borrower in 2026 may arrive with a convincing driver’s license generated by image models, employment verification supported by AI-written HR correspondence, and a live onboarding video featuring a deepfake face synchronized with cloned speech patterns. Some fraud operations are reportedly generating entire digital footprints.
Across credit unions, for example, PYMNTS Intelligence research found that fraud now occurs across the full member life cycle, from account opening and onboarding to authentication and transaction activity. Credit unions must now defend every interaction point rather than a single stage, and 77% of them have experienced unauthorized network access in the past year.
Traditional fraud detection relies heavily on identifying outliers, like sudden spikes in credit activity, inconsistent identity data, abnormal transaction patterns or behavioral irregularities during onboarding.
“If you are sitting in a café or restaurant, people’s behavior typically is fairly similar,” James Mirfin, senior vice president, Global Head of Risk and Security Intelligence Solutions at Visa, told PYMNTS in March. “But you can spot someone that looks a bit nervous or twitchy. It’s the same in banking. Historically, bank tellers were looking for anomalous behavior.”
However, AI systems trained on large datasets can now generate borrowers engineered specifically to avoid those signals.
“If a human can do it, we are now at a stage where the machines can do it in plausible ways,” Adam Hiatt, vice president of fraud strategy at Spreedly, said. “It’s an arms race.”
That creates a paradox for lenders. The same machine-learning systems built to reduce human bias and expand access to credit may also be unusually susceptible to applicants engineered by generative AI. The threat becomes even more acute for credit unions and FinTech platforms that compete on frictionless customer acquisition. Many digital lenders have optimized onboarding around convenience, with instant approvals, remote verification, minimal documentation and reduced human intervention. Every additional verification step risks increasing abandonment rates and reducing conversion.
Read also: Cybersecurity’s Hottest New Job Is Negotiating With Hackers
Deepfakes Turn Identity Verification Into Theater
Synthetic borrowers are exploiting the same seamless digital experiences that legitimate consumers increasingly expect. Advances in generative video now allow fraudsters to produce realistic facial movements, synchronized speech and convincing real-time interactions. Voice cloning systems can replicate tone, cadence and emotional inflection from relatively small audio samples. Combined with stolen or fabricated identity documents, these tools enable entirely synthetic applicants to pass forms of verification that were considered highly secure only a few years ago.
“With synthetic voices now, bad actors can scrape voices off the internet via social media channels, take a snippet of that voice and then create an entire script based on that person’s voice,” Elizabeth Wadsworth, vice president of Decision Intelligence and Transformation at Velera, told PYMNTS in March.
“We’re in a space where it’s hitting all sides,” she added.
The PYMNTS Intelligence report “The AI MonitorEdge Report: COOs Leverage GenAI to Reduce Data Security Losses” showed that 55% of companies are employing AI-powered cybersecurity measures. But a lender may spend millions of dollars developing anti-fraud infrastructure, while open-source generative AI tools lower the cost of attack for criminal organizations. Fraud rings no longer require elite technical expertise; many capabilities are becoming commercially accessible.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
