AI-Powered Fraud Schemes Target Bank Onboarding Systems With Synthetic Identity Attacks

A new fraud intelligence report indicates that fraudsters are using generative AI to bypass the digital onboarding systems banks use to approve new customer accounts.

AI-generated identities can pass automated identity checks when opening a new account, according to researchers at DuckDuckGoose.

The report states that fraudsters will be able to generate convincing digital personas at a faster rate than the bank’s ability to update its verification systems. 

This creates a serious problem for credit card issuers because most rely on automated onboarding tools that perform identity verification processes via document upload, facial scan, and liveness check.

That onboarding step has effectively become the front door to the credit ecosystems as more credit card applications move to fully digital, human-free approval systems.

“Identity verification systems are entering a structural risk phase as generative AI enables synthetic identities to be created faster than verification infrastructure can adapt,” according to the report.

Synthetic identity fraud is among the hardest financial crimes to detect. In synthetic identity fraud, criminals fabricate a new identity for themselves using a combination of false information and legitimate personal data. 

Once the fraudulent identity has passed through verification, the fraudster opens an account with a lender and creates a credit history. As the credit history grows, so do the fraudster’s credit limits enabling them to open accounts with multiple other lenders.

The fraudster then carries out what is referred to as a “bust-out,” in which they rapidly borrow against their credit limits and disappear with the borrowed funds.

The use of Generative AI will make it much easier for fraudsters to carry out their scheme on a larger scale. Tools available online now generate fake facial images and video images (liveness), and provide forgeries of identification documents such as passports and drivers’ licenses.

According to the DuckDuckGoose report, more than 55 synthetic media creation tools were released in the fourth quarter of 2025 — approximately one new tool every 1.6 days.

Since early 2024, researchers have also reported a rapid increase — over 1,000% in fact — in image-to-video creation tools that can produce a short video that mimics human motion. This potentially allows AI-created identities to successfully pass through selfie-based identity verification systems.

As you can see from the prior statements, synthetic identity fraud has become a major financial concern, with more than $35 billion lost globally in recent years.

The increasing sophistication of AI-generated identities could force credit card issuers to rethink their digital onboarding process.

Identity verification has traditionally relied on government identification documents, facial recognition, and checking databases for information. Generative AI is rapidly improving to create similar signals.

Credit card issuers are likely to react to this threat with the addition of new security layers to their onboarding processes, including behavioral biometric tools, device fingerprinting technologies, and fraud intelligence sharing networks.

AI can potentially become an additional layer of defense for these organizations. In particular, machine learning algorithms can identify subtle anomalies in an individual’s identity information or device activity when they apply for credit.

The issue facing credit card issuers will be balancing speed and convenience with the need to prevent AI-generated identities from being accepted into the credit system.